CIRCULATE Galaxy Nakamura et al 2024

library(swimplot) library(coxphf) library(grid) library(gtable) library(readr) library(mosaic) library(dplyr) library(survival) library(survminer) library(ggplot2) library(scales) library(ggthemes) library(tidyverse) library(gtsummary) library(flextable) library(parameters) library(car) library(grid) library(ComplexHeatmap) library(readxl) library(janitor) library(rms) library(DT)

#Demographics Table

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]

circ_data_subset <- circ_data %>%
  select(
    Age,
    Gender,
    ECOG,
    PrimSite,
    pT,
    pN,
    Stage,
    NAC,
    ACT,
    BRAF.V600E,
    RAS,
    MSI,
    RFS.Event,
    OS.months) %>%
  mutate(
    Age = as.numeric(Age),
    Gender = factor(Gender, levels = c("Male", "Female")),
    ECOG = factor(ECOG, levels = c(0, 1)),
    PrimSite = factor(PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum")),
    pT = factor(pT, levels = c("T1-T2", "T3-T4")),
    pN = factor(pN, levels = c("N0", "N1-N2")),
    Stage = factor(Stage, levels = c("I","II", "III", "IV")),
    NAC = factor(NAC, levels = c("TRUE", "FALSE"), labels = c("Neoadjuvant Chemotherapy", "Upfront Surgery")),
    ACT = factor(ACT, levels = c("TRUE", "FALSE"), labels = c("Adjuvant Chemotherapy", "Observation")),
    BRAF.V600E = factor(BRAF.V600E, levels = c("WT", "MUT"), labels = c("BRAF wt", "BRAF V600E")),
    RAS = factor(RAS, levels = c("WT", "MUT"), labels = c("RAS wt", "RAS mut")),
    MSI = factor(MSI, levels = c("MSS", "MSI-High")),
    RFS.Event = factor(RFS.Event, levels = c("TRUE", "FALSE"), labels = c("Recurrence", "No Recurrence")),
    OS.months = as.numeric(OS.months))
table1 <- circ_data_subset %>%
  tbl_summary(
    statistic = list(
      all_continuous() ~ "{median} ({min} - {max})",
      all_categorical() ~ "{n} ({p}%)")) %>%
  bold_labels()
table1

Characteristic	N = 2,2401
Age	69 (28 - 95)
Gender
Male	1,149 (51%)
Female	1,091 (49%)
ECOG
0	2,046 (91%)
1	194 (8.7%)
PrimSite
Right-sided colon	863 (39%)
Left-sided colon	1,377 (61%)
Rectum	0 (0%)
pT
T1-T2	317 (16%)
T3-T4	1,630 (84%)
Unknown	293
pN
N0	922 (47%)
N1-N2	1,025 (53%)
Unknown	293
Stage
I	234 (10%)
II	652 (29%)
III	936 (42%)
IV	418 (19%)
NAC
Neoadjuvant Chemotherapy	218 (9.7%)
Upfront Surgery	2,022 (90%)
ACT
Adjuvant Chemotherapy	946 (42%)
Observation	1,294 (58%)
BRAF.V600E
BRAF wt	2,062 (92%)
BRAF V600E	178 (7.9%)
RAS
RAS wt	1,303 (58%)
RAS mut	937 (42%)
MSI
MSS	2,025 (90%)
MSI-High	215 (9.6%)
RFS.Event
Recurrence	500 (22%)
No Recurrence	1,740 (78%)
OS.months	23 (2 - 49)
1 Median (Range); n (%)

fit1 <- as_flex_table(
  table1,
  include = everything(),
  return_calls = FALSE,
  strip_md_bold = TRUE)
fit1

Characteristic	N = 2,2401
Age	69 (28 - 95)
Gender
Male	1,149 (51%)
Female	1,091 (49%)
ECOG
0	2,046 (91%)
1	194 (8.7%)
PrimSite
Right-sided colon	863 (39%)
Left-sided colon	1,377 (61%)
Rectum	0 (0%)
pT
T1-T2	317 (16%)
T3-T4	1,630 (84%)
Unknown	293
pN
N0	922 (47%)
N1-N2	1,025 (53%)
Unknown	293
Stage
I	234 (10%)
II	652 (29%)
III	936 (42%)
IV	418 (19%)
NAC
Neoadjuvant Chemotherapy	218 (9.7%)
Upfront Surgery	2,022 (90%)
ACT
Adjuvant Chemotherapy	946 (42%)
Observation	1,294 (58%)
BRAF.V600E
BRAF wt	2,062 (92%)
BRAF V600E	178 (7.9%)
RAS
RAS wt	1,303 (58%)
RAS mut	937 (42%)
MSI
MSS	2,025 (90%)
MSI-High	215 (9.6%)
RFS.Event
Recurrence	500 (22%)
No Recurrence	1,740 (78%)
OS.months	23 (2 - 49)
1Median (Range); n (%)

save_as_docx(fit1, path= "~/Downloads/table1.docx")

#ctDNA Detection Rates by Window and Stages

#ctDNA at Baseline
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data$ctDNA.Baseline <- factor(circ_data$ctDNA.Baseline, levels=c("NEGATIVE","POSITIVE"))
circ_data <- subset(circ_data, ctDNA.Baseline %in% c("NEGATIVE", "POSITIVE"))
circ_data$Stage <- factor(circ_data$Stage, levels=c("I","II", "III","IV"))
positive_counts_by_stage <- aggregate(circ_data$ctDNA.Baseline == "POSITIVE", by=list(circ_data$Stage), FUN=sum)
total_counts_by_stage <- aggregate(circ_data$ctDNA.Baseline, by=list(circ_data$Stage), FUN=length)
combined_data <- data.frame(
  Stage = total_counts_by_stage$Group.1,
  Total_Count = total_counts_by_stage$x,
  Positive_Count = positive_counts_by_stage$x,
  Rate = (positive_counts_by_stage$x / total_counts_by_stage$x) * 100  # Convert to percentage
)
combined_data$Rate <- sprintf("%.2f%%", combined_data$Rate)
overall_total_count <- nrow(circ_data)
overall_positive_count <- nrow(circ_data[circ_data$ctDNA.Baseline == "POSITIVE",])
overall_positivity_rate <- (overall_positive_count / overall_total_count) * 100  # Convert to percentage
overall_row <- data.frame(
  Stage = "Overall",
  Total_Count = overall_total_count,
  Positive_Count = overall_positive_count,
  Rate = sprintf("%.2f%%", overall_positivity_rate)
)
combined_data <- rbind(combined_data, overall_row)
print(combined_data)

#ctDNA at MRD Window
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
circ_data$Stage <- factor(circ_data$Stage, levels=c("I","II", "III","IV"))
positive_counts_by_stage <- aggregate(circ_data$ctDNA.MRD == "POSITIVE", by=list(circ_data$Stage), FUN=sum)
total_counts_by_stage <- aggregate(circ_data$ctDNA.MRD, by=list(circ_data$Stage), FUN=length)
combined_data <- data.frame(
  Stage = total_counts_by_stage$Group.1,
  Total_Count = total_counts_by_stage$x,
  Positive_Count = positive_counts_by_stage$x,
  Rate = (positive_counts_by_stage$x / total_counts_by_stage$x) * 100  # Convert to percentage
)
combined_data$Rate <- sprintf("%.2f%%", combined_data$Rate)
overall_total_count <- nrow(circ_data)
overall_positive_count <- nrow(circ_data[circ_data$ctDNA.MRD == "POSITIVE",])
overall_positivity_rate <- (overall_positive_count / overall_total_count) * 100  # Convert to percentage
overall_row <- data.frame(
  Stage = "Overall",
  Total_Count = overall_total_count,
  Positive_Count = overall_positive_count,
  Rate = sprintf("%.2f%%", overall_positivity_rate)
)
combined_data <- rbind(combined_data, overall_row)
print(combined_data)

#ctDNA at Surveillance Window
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data$ctDNA.Surveillance <- factor(circ_data$ctDNA.Surveillance, levels=c("NEGATIVE","POSITIVE"))
circ_data <- subset(circ_data, ctDNA.Surveillance %in% c("NEGATIVE", "POSITIVE"))
circ_data$Stage <- factor(circ_data$Stage, levels=c("I","II", "III","IV"))
positive_counts_by_stage <- aggregate(circ_data$ctDNA.Surveillance == "POSITIVE", by=list(circ_data$Stage), FUN=sum)
total_counts_by_stage <- aggregate(circ_data$ctDNA.Surveillance, by=list(circ_data$Stage), FUN=length)
combined_data <- data.frame(
  Stage = total_counts_by_stage$Group.1,
  Total_Count = total_counts_by_stage$x,
  Positive_Count = positive_counts_by_stage$x,
  Rate = (positive_counts_by_stage$x / total_counts_by_stage$x) * 100  # Convert to percentage
)
combined_data$Rate <- sprintf("%.2f%%", combined_data$Rate)
overall_total_count <- nrow(circ_data)
overall_positive_count <- nrow(circ_data[circ_data$ctDNA.Surveillance == "POSITIVE",])
overall_positivity_rate <- (overall_positive_count / overall_total_count) * 100  # Convert to percentage
overall_row <- data.frame(
  Stage = "Overall",
  Total_Count = overall_total_count,
  Positive_Count = overall_positive_count,
  Rate = sprintf("%.2f%%", overall_positivity_rate)
)
combined_data <- rbind(combined_data, overall_row)
print(combined_data)

#DFS by ctDNA at the MRD Window - All stages Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   1 observation deleted due to missingness 
                      n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 1773    233     NA      NA      NA
ctDNA.MRD=POSITIVE  336    263   5.34    4.83     6.7

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | All stages", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24, 30, 36))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

1 observation deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24    625     224    0.851 0.00949        0.832        0.869
   30    353       6    0.841 0.01025        0.820        0.860
   36    131       2    0.835 0.01101        0.812        0.856

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24     36     258    0.206  0.0236        0.161        0.254
   30     21       3    0.185  0.0242        0.140        0.234
   36     10       2    0.167  0.0250        0.121        0.219

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 2109, number of events= 496 
   (1 observation deleted due to missingness)

                      coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  2.48392  11.98819  0.09162 27.11   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     11.99    0.08342     10.02     14.35

Concordance= 0.738  (se = 0.01 )
Likelihood ratio test= 631.6  on 1 df,   p=<2e-16
Wald test            = 734.9  on 1 df,   p=<2e-16
Score (logrank) test = 1164  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 11.99 (10.02-14.35); p = 0"

#DFS by ctDNA at the MRD Window - Stage I Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[!(circ_data$Stage %in% c("II", "III", "IV")),]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 226      7     NA      NA      NA
ctDNA.MRD=POSITIVE   2      2   15.3   0.526      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | Stage I", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      67.0000       7.0000       0.9556       0.0176       0.9043       0.9797 

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
    24.00000      1.00000      1.00000      0.50000      0.35355      0.00598      0.91041 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 228, number of events= 9 

                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  3.5700   35.5148   0.8291 4.306 1.66e-05 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     35.51    0.02816     6.993     180.4

Concordance= 0.587  (se = 0.069 )
Likelihood ratio test= 9.72  on 1 df,   p=0.002
Wald test            = 18.54  on 1 df,   p=2e-05
Score (logrank) test = 47.16  on 1 df,   p=7e-12

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 35.51 (6.99-180.35); p = 0"

#DFS by ctDNA at the MRD Window - Stage II Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "III", "IV")),]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 584     30     NA      NA      NA
ctDNA.MRD=POSITIVE  45     30   7.75    5.45      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | Stage II", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     234.0000      29.0000       0.9413       0.0108       0.9159       0.9592 

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       6.0000      29.0000       0.3250       0.0749       0.1864       0.4714 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 629, number of events= 60 

                     coef exp(coef) se(coef)    z Pr(>|z|)    
ctDNA.MRDPOSITIVE  3.1738   23.8977   0.2623 12.1   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE      23.9    0.04184     14.29     39.96

Concordance= 0.745  (se = 0.031 )
Likelihood ratio test= 110.6  on 1 df,   p=<2e-16
Wald test            = 146.4  on 1 df,   p=<2e-16
Score (logrank) test = 310.6  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 23.9 (14.29-39.96); p = 0"

#DFS by ctDNA at the MRD Window - Stage II & T3N0/T4N0 Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "III", "IV")),]
circ_data <- circ_data[circ_data$StageII.Group!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Stage.II.Risk <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Stage.II.TNM = case_when(
    ctDNA.MRD == "NEGATIVE" & StageII.Group == "T3N0" ~ 1,
    ctDNA.MRD == "POSITIVE" & StageII.Group == "T3N0" ~ 2,
    ctDNA.MRD == "NEGATIVE" & StageII.Group == "T4N0" ~ 3,
    ctDNA.MRD == "POSITIVE" & StageII.Group == "T4N0" ~ 4
  ))

circ_data <- circ_data[circ_data$ctDNA.Stage.II.TNM!="",]
survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.Stage.II.TNM, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.Stage.II.TNM, data = circ_data)

   17 observations deleted due to missingness 
                       n events median 0.95LCL 0.95UCL
ctDNA.Stage.II.TNM=1 476     18     NA      NA      NA
ctDNA.Stage.II.TNM=2  29     18  10.74    6.14      NA
ctDNA.Stage.II.TNM=3  93     11     NA      NA      NA
ctDNA.Stage.II.TNM=4  14     10   5.22    4.37      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Stage.II.TNM) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Stage.II.TNM, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","green","purple", "red"), title="DFS - ctDNA MRD & Stage II TNM", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA(-) & T3N0", "ctDNA(+) & T3N0", "ctDNA(-) & T4N0", "ctDNA(+) & T4N0"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.Stage.II.TNM, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

17 observations deleted due to missingness 
                ctDNA.Stage.II.TNM=1 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     199.0000      17.0000       0.9561       0.0107       0.9295       0.9729 

                ctDNA.Stage.II.TNM=2 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       4.0000      17.0000       0.3638       0.0996       0.1793       0.5516 

                ctDNA.Stage.II.TNM=3 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      26.0000      11.0000       0.8680       0.0375       0.7730       0.9252 

                ctDNA.Stage.II.TNM=4 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       2.0000      10.0000       0.2857       0.1207       0.0883       0.5237 

circ_data$ctDNA.Stage.II.TNM <- factor(circ_data$ctDNA.Stage.II.TNM, levels=c("1","2","3","4"), labels = c("ctDNA(-) & T3N0", "ctDNA(+) & T3N0", "ctDNA(-) & T4N0", "ctDNA(+) & T4N0"))
cox_fit <- coxph(surv_object ~ ctDNA.Stage.II.TNM, data=circ_data) 
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Stage.II.TNM, data = circ_data)

  n= 612, number of events= 57 
   (17 observations deleted due to missingness)

                                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.Stage.II.TNMctDNA(+) & T3N0  3.3181   27.6092   0.3364 9.864  < 2e-16 ***
ctDNA.Stage.II.TNMctDNA(-) & T4N0  1.1962    3.3077   0.3829 3.124  0.00178 ** 
ctDNA.Stage.II.TNMctDNA(+) & T4N0  3.6897   40.0340   0.3977 9.277  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.Stage.II.TNMctDNA(+) & T3N0    27.609    0.03622    14.279    53.382
ctDNA.Stage.II.TNMctDNA(-) & T4N0     3.308    0.30233     1.562     7.006
ctDNA.Stage.II.TNMctDNA(+) & T4N0    40.034    0.02498    18.360    87.295

Concordance= 0.798  (se = 0.032 )
Likelihood ratio test= 110.8  on 3 df,   p=<2e-16
Wald test            = 135  on 3 df,   p=<2e-16
Score (logrank) test = 297.7  on 3 df,   p=<2e-16

#Repeat analysis to compare ctDNA MRD (-) vs (+) in T4N0
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "III", "IV")),]
circ_data <- circ_data[circ_data$StageII.Group!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Stage.II.Risk <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Stage.II.TNM = case_when(
    ctDNA.MRD == "NEGATIVE" & StageII.Group == "T3N0" ~ 1,
    ctDNA.MRD == "POSITIVE" & StageII.Group == "T3N0" ~ 2,
    ctDNA.MRD == "NEGATIVE" & StageII.Group == "T4N0" ~ 3,
    ctDNA.MRD == "POSITIVE" & StageII.Group == "T4N0" ~ 4
  ))

circ_data <- circ_data[circ_data$ctDNA.Stage.II.TNM!="",]
survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.Stage.II.TNM, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.Stage.II.TNM, data = circ_data)

   17 observations deleted due to missingness 
                       n events median 0.95LCL 0.95UCL
ctDNA.Stage.II.TNM=1 476     18     NA      NA      NA
ctDNA.Stage.II.TNM=2  29     18  10.74    6.14      NA
ctDNA.Stage.II.TNM=3  93     11     NA      NA      NA
ctDNA.Stage.II.TNM=4  14     10   5.22    4.37      NA

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Stage.II.TNM, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","green","purple", "red"), title="DFS - ctDNA MRD & Stage II TNM", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA(-) & T3N0", "ctDNA(+) & T3N0", "ctDNA(-) & T4N0", "ctDNA(+) & T4N0"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.Stage.II.TNM, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

17 observations deleted due to missingness 
                ctDNA.Stage.II.TNM=1 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     199.0000      17.0000       0.9561       0.0107       0.9295       0.9729 

                ctDNA.Stage.II.TNM=2 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       4.0000      17.0000       0.3638       0.0996       0.1793       0.5516 

                ctDNA.Stage.II.TNM=3 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      26.0000      11.0000       0.8680       0.0375       0.7730       0.9252 

                ctDNA.Stage.II.TNM=4 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       2.0000      10.0000       0.2857       0.1207       0.0883       0.5237 

circ_data$ctDNA.Stage.II.TNM <- factor(circ_data$ctDNA.Stage.II.TNM, levels=c("2","4","1","3"))
cox_fit <- coxph(surv_object ~ ctDNA.Stage.II.TNM, data=circ_data) 
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Stage.II.TNM, data = circ_data)

  n= 612, number of events= 57 
   (17 observations deleted due to missingness)

                        coef exp(coef) se(coef)      z Pr(>|z|)    
ctDNA.Stage.II.TNM4  0.37158   1.45003  0.39523  0.940    0.347    
ctDNA.Stage.II.TNM1 -3.31815   0.03622  0.33640 -9.864  < 2e-16 ***
ctDNA.Stage.II.TNM3 -2.12190   0.11980  0.38492 -5.513 3.54e-08 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                    exp(coef) exp(-coef) lower .95 upper .95
ctDNA.Stage.II.TNM4   1.45003     0.6896   0.66828   3.14625
ctDNA.Stage.II.TNM1   0.03622    27.6092   0.01873   0.07003
ctDNA.Stage.II.TNM3   0.11980     8.3470   0.05634   0.25475

Concordance= 0.798  (se = 0.032 )
Likelihood ratio test= 110.8  on 3 df,   p=<2e-16
Wald test            = 135  on 3 df,   p=<2e-16
Score (logrank) test = 297.7  on 3 df,   p=<2e-16

#DFS by ctDNA at the MRD Window - Stage III Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "IV")),]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   1 observation deleted due to missingness 
                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 683     82     NA      NA      NA
ctDNA.MRD=POSITIVE 162    117   9.48    7.16    11.7

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | Stage III", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

1 observation deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     242.0000      78.0000       0.8600       0.0152       0.8272       0.8870 

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000       22.000      115.000        0.259        0.037        0.190        0.334 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 845, number of events= 199 
   (1 observation deleted due to missingness)

                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  2.3582   10.5722   0.1459 16.16   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     10.57    0.09459     7.942     14.07

Concordance= 0.752  (se = 0.016 )
Likelihood ratio test= 245  on 1 df,   p=<2e-16
Wald test            = 261.2  on 1 df,   p=<2e-16
Score (logrank) test = 399.4  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 10.57 (7.94-14.07); p = 0"

#DFS by ctDNA at the MRD Window - High Risk Stage II Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$Risk.StageII==TRUE,]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   1481 observations deleted due to missingness 
                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 475     24     NA      NA      NA
ctDNA.MRD=POSITIVE  42     28   7.56    4.99      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | High Risk Stage II", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

1481 observations deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000      193.000       23.000        0.942        0.012        0.914        0.962 

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000        6.000       27.000        0.337        0.076        0.195        0.484 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 517, number of events= 52 
   (1481 observations deleted due to missingness)

                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  3.2102   24.7836   0.2831 11.34   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     24.78    0.04035     14.23     43.16

Concordance= 0.764  (se = 0.033 )
Likelihood ratio test= 102.4  on 1 df,   p=<2e-16
Wald test            = 128.6  on 1 df,   p=<2e-16
Score (logrank) test = 275.5  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 24.78 (14.23-43.16); p = 0"

#DFS by ctDNA at the MRD Window - High Risk Stage III Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$Risk.StageIII==TRUE,]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   1265 observations deleted due to missingness 
                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 383     56     NA      NA      NA
ctDNA.MRD=POSITIVE 105     79   10.1    7.66      14

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | High Risk Stage III", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

1265 observations deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     130.0000      53.0000       0.8322       0.0219       0.7842       0.8705 

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      13.0000      77.0000       0.2305       0.0443       0.1500       0.3214 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 488, number of events= 135 
   (1265 observations deleted due to missingness)

                    coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE 2.2154    9.1654   0.1775 12.48   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     9.165     0.1091     6.472     12.98

Concordance= 0.74  (se = 0.019 )
Likelihood ratio test= 147.9  on 1 df,   p=<2e-16
Wald test            = 155.7  on 1 df,   p=<2e-16
Score (logrank) test = 226.5  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 9.17 (6.47-12.98); p = 0"

#DFS by ctDNA at the MRD Window - Stage IV Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 280    114     NA   26.91      NA
ctDNA.MRD=POSITIVE 127    114   2.83    2.17    4.21

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | Stage IV", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      82.0000     110.0000       0.5748       0.0319       0.5097       0.6344 

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       7.0000     113.0000       0.0924       0.0274       0.0479       0.1548 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 407, number of events= 228 

                    coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE 1.7624    5.8266   0.1384 12.73   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     5.827     0.1716     4.442     7.642

Concordance= 0.695  (se = 0.013 )
Likelihood ratio test= 148.1  on 1 df,   p=<2e-16
Wald test            = 162.2  on 1 df,   p=<2e-16
Score (logrank) test = 200.2  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 5.83 (4.44-7.64); p = 0"

#OS by ctDNA at the MRD Window - All stages Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   1 observation deleted due to missingness 
                      n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 1773     36     NA      NA      NA
ctDNA.MRD=POSITIVE  336     52   43.4      NA      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="OS - ctDNA MRD window | All stages", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24, 30, 36))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

1 observation deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24    825      18    0.985 0.00349        0.977        0.991
   30    497      13    0.968 0.00593        0.954        0.978
   36    185       4    0.960 0.00722        0.943        0.972

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24    119      37    0.837  0.0258        0.778        0.881
   30     73       9    0.769  0.0323        0.698        0.825
   36     24       4    0.718  0.0388        0.634        0.786

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 2109, number of events= 88 
   (1 observation deleted due to missingness)

                   coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE 2.271     9.685    0.217 10.46   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     9.685     0.1033      6.33     14.82

Concordance= 0.754  (se = 0.027 )
Likelihood ratio test= 103.2  on 1 df,   p=<2e-16
Wald test            = 109.5  on 1 df,   p=<2e-16
Score (logrank) test = 165.2  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 9.68 (6.33-14.82); p = 0"

#Multivariate cox regression at MRD Window for DFS - All stages Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"), labels = c("Negative", "Positive"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", ">70"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels = c("0", "1"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-High"), labels = c("MSS", "MSI-High"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"), labels = c("Wild-Type", "V600E"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"), labels = c("Wild-Type", "Mutant"))
surv_object <- Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ctDNA.MRD + Gender + Age.Group + PrimSite + ECOG + pT + pN + MSI + BRAF.V600E + RAS, data=circ_data) 
ggforest(cox_fit, data = circ_data, main = "Multivariate Regression Model for DFS - All Stages", refLabel = "Reference Group")

test.ph <- cox.zph(cox_fit)

#Multivariate cox regression at MRD Window for OS - All stages Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"), labels = c("Negative", "Positive"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", ">70"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels = c("0", "1"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-High"), labels = c("MSS", "MSI-High"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"), labels = c("Wild-Type", "V600E"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"), labels = c("Wild-Type", "Mutant"))
surv_object <- Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event) 
cox_fit <- coxph(surv_object ~ ctDNA.MRD + Gender + Age.Group + PrimSite + ECOG + pT + pN + MSI + BRAF.V600E + RAS, data=circ_data) 
ggforest(cox_fit, data = circ_data, main = "Multivariate Regression Model for OS - All Stages", refLabel = "Reference Group")

test.ph <- cox.zph(cox_fit)

#DFS by ACT treatment in MRD negative - High Risk Stage II/III

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data <- circ_data[circ_data$HighRisk.Stage=="TRUE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

   15 observations deleted due to missingness 
            n events median 0.95LCL 0.95UCL
ACT=FALSE 586     50     NA      NA      NA
ACT=TRUE  571     55     NA      NA      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Negative ACT vs Observation | High Risk Stage II/III", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

15 observations deleted due to missingness 
                ACT=FALSE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000      215.000       49.000        0.899        0.014        0.868        0.923 

                ACT=TRUE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     216.0000      51.0000       0.8911       0.0148       0.8581       0.9168 

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 1157, number of events= 105 
   (15 observations deleted due to missingness)

            coef exp(coef) se(coef)      z Pr(>|z|)
ACTFALSE -0.1149    0.8915   0.1954 -0.588    0.557

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE    0.8915      1.122    0.6078     1.307

Concordance= 0.508  (se = 0.025 )
Likelihood ratio test= 0.35  on 1 df,   p=0.6
Wald test            = 0.35  on 1 df,   p=0.6
Score (logrank) test = 0.35  on 1 df,   p=0.6

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 0.89 (0.61-1.31); p = 0.557"

#Adjusted HR "ACT vs no ACT" - age, gender, ECOG and pathological stage
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data <- circ_data[circ_data$HighRisk.Stage=="TRUE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + Stage + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + Stage + 
    ECOG, data = circ_data)

  n= 1157, number of events= 105 
   (15 observations deleted due to missingness)

                coef exp(coef) se(coef)      z Pr(>|z|)    
ACTFALSE      0.3623    1.4367   0.2145  1.689   0.0911 .  
GenderMale    0.1477    1.1591   0.1960  0.753   0.4512    
Age.Group≥70 -0.3075    0.7353   0.2067 -1.487   0.1369    
StageIII      1.0528    2.8656   0.2528  4.164 3.13e-05 ***
ECOG1         0.2435    1.2756   0.3168  0.769   0.4422    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        1.4367     0.6961    0.9436     2.187
GenderMale      1.1591     0.8627    0.7894     1.702
Age.Group≥70    0.7353     1.3600    0.4903     1.103
StageIII        2.8656     0.3490    1.7458     4.704
ECOG1           1.2756     0.7839    0.6856     2.373

Concordance= 0.629  (se = 0.026 )
Likelihood ratio test= 23.38  on 5 df,   p=3e-04
Wald test            = 21.22  on 5 df,   p=7e-04
Score (logrank) test = 22.35  on 5 df,   p=4e-04

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data <- circ_data[circ_data$HighRisk.Stage=="TRUE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + Stage + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + Stage + 
    ECOG, data = circ_data)

  n= 1157, number of events= 105 
   (15 observations deleted due to missingness)

                coef exp(coef) se(coef)      z Pr(>|z|)    
ACTTRUE      -0.3623    0.6961   0.2145 -1.689   0.0911 .  
GenderMale    0.1477    1.1591   0.1960  0.753   0.4512    
Age.Group≥70 -0.3075    0.7353   0.2067 -1.487   0.1369    
StageIII      1.0528    2.8656   0.2528  4.164 3.13e-05 ***
ECOG1         0.2435    1.2756   0.3168  0.769   0.4422    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.6961     1.4367    0.4572     1.060
GenderMale      1.1591     0.8627    0.7894     1.702
Age.Group≥70    0.7353     1.3600    0.4903     1.103
StageIII        2.8656     0.3490    1.7458     4.704
ECOG1           1.2756     0.7839    0.6856     2.373

Concordance= 0.629  (se = 0.026 )
Likelihood ratio test= 23.38  on 5 df,   p=3e-04
Wald test            = 21.22  on 5 df,   p=7e-04
Score (logrank) test = 22.35  on 5 df,   p=4e-04

#DFS by ACT treatment in MRD positive - High Risk Stage II/III

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data <- circ_data[circ_data$HighRisk.Stage=="TRUE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

   1 observation deleted due to missingness 
            n events median 0.95LCL 0.95UCL
ACT=FALSE  47     45   3.55    3.16    3.95
ACT=TRUE  145     88  12.06    9.30   18.57

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Positive ACT vs Observation | High Risk Stage II/III", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

1 observation deleted due to missingness 
                ACT=FALSE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
    24.00000      1.00000     44.00000      0.02837      0.02746      0.00232      0.12350 

                ACT=TRUE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      25.0000      87.0000       0.3583       0.0435       0.2741       0.4432 

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 192, number of events= 133 
   (1 observation deleted due to missingness)

           coef exp(coef) se(coef)     z Pr(>|z|)    
ACTFALSE 1.4203    4.1382   0.1901 7.472 7.91e-14 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     4.138     0.2417     2.851     6.006

Concordance= 0.634  (se = 0.019 )
Likelihood ratio test= 46.68  on 1 df,   p=8e-12
Wald test            = 55.83  on 1 df,   p=8e-14
Score (logrank) test = 64.71  on 1 df,   p=9e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 4.14 (2.85-6.01); p = 0"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI and pathological stage
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data <- circ_data[circ_data$HighRisk.Stage=="TRUE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + Stage + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + Stage + 
    ECOG, data = circ_data)

  n= 192, number of events= 133 
   (1 observation deleted due to missingness)

                 coef exp(coef) se(coef)      z Pr(>|z|)    
ACTFALSE      1.46226   4.31571  0.20651  7.081 1.43e-12 ***
GenderMale   -0.06402   0.93799  0.18183 -0.352    0.725    
Age.Group≥70  0.03736   1.03807  0.18637  0.200    0.841    
StageIII      0.31989   1.37697  0.23571  1.357    0.175    
ECOG1         0.05652   1.05814  0.28089  0.201    0.841    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE         4.316     0.2317    2.8792     6.469
GenderMale       0.938     1.0661    0.6568     1.340
Age.Group≥70     1.038     0.9633    0.7204     1.496
StageIII         1.377     0.7262    0.8675     2.186
ECOG1            1.058     0.9451    0.6102     1.835

Concordance= 0.644  (se = 0.026 )
Likelihood ratio test= 49.19  on 5 df,   p=2e-09
Wald test            = 58.77  on 5 df,   p=2e-11
Score (logrank) test = 67.68  on 5 df,   p=3e-13

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data <- circ_data[circ_data$HighRisk.Stage=="TRUE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + Stage + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + Stage + 
    ECOG, data = circ_data)

  n= 192, number of events= 133 
   (1 observation deleted due to missingness)

                 coef exp(coef) se(coef)      z Pr(>|z|)    
ACTTRUE      -1.46226   0.23171  0.20651 -7.081 1.43e-12 ***
GenderMale   -0.06402   0.93799  0.18183 -0.352    0.725    
Age.Group≥70  0.03736   1.03807  0.18637  0.200    0.841    
StageIII      0.31989   1.37697  0.23571  1.357    0.175    
ECOG1         0.05652   1.05814  0.28089  0.201    0.841    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.2317     4.3157    0.1546    0.3473
GenderMale      0.9380     1.0661    0.6568    1.3396
Age.Group≥70    1.0381     0.9633    0.7204    1.4958
StageIII        1.3770     0.7262    0.8675    2.1855
ECOG1           1.0581     0.9451    0.6102    1.8350

Concordance= 0.644  (se = 0.026 )
Likelihood ratio test= 49.19  on 5 df,   p=2e-09
Wald test            = 58.77  on 5 df,   p=2e-11
Score (logrank) test = 67.68  on 5 df,   p=3e-13

#DFS by ACT treatment in MRD negative - High Risk Stage II

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$Risk.StageII==TRUE,]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

   1588 observations deleted due to missingness 
            n events median 0.95LCL 0.95UCL
ACT=FALSE 373     21     NA      NA      NA
ACT=TRUE  102      3     NA      NA      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Negative ACT vs Observation | High Risk Stage II", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

1588 observations deleted due to missingness 
                ACT=FALSE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000      152.000       20.000        0.937        0.014        0.903        0.959 

                ACT=TRUE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      38.0000       3.0000       0.9634       0.0211       0.8890       0.9883 

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 475, number of events= 24 
   (1588 observations deleted due to missingness)

           coef exp(coef) se(coef)     z Pr(>|z|)
ACTFALSE 0.6344    1.8860   0.6173 1.028    0.304

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     1.886     0.5302    0.5625     6.323

Concordance= 0.544  (se = 0.035 )
Likelihood ratio test= 1.23  on 1 df,   p=0.3
Wald test            = 1.06  on 1 df,   p=0.3
Score (logrank) test = 1.09  on 1 df,   p=0.3

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 1.89 (0.56-6.32); p = 0.304"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$Risk.StageII==TRUE,]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 475, number of events= 24 
   (1588 observations deleted due to missingness)

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTFALSE      0.7519    2.1211   0.6266  1.200   0.2301  
GenderMale   -0.1514    0.8595   0.4160 -0.364   0.7159  
Age.Group≥70 -0.8105    0.4446   0.4420 -1.834   0.0667 .
ECOG1         0.5506    1.7343   0.5794  0.950   0.3419  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        2.1211     0.4715    0.6212     7.243
GenderMale      0.8595     1.1634    0.3803     1.943
Age.Group≥70    0.4446     2.2490    0.1870     1.057
ECOG1           1.7343     0.5766    0.5571     5.399

Concordance= 0.629  (se = 0.06 )
Likelihood ratio test= 4.98  on 4 df,   p=0.3
Wald test            = 4.66  on 4 df,   p=0.3
Score (logrank) test = 4.79  on 4 df,   p=0.3

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$Risk.StageII==TRUE,]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))

circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 475, number of events= 24 
   (1588 observations deleted due to missingness)

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTTRUE      -0.7519    0.4715   0.6266 -1.200   0.2301  
GenderMale   -0.1514    0.8595   0.4160 -0.364   0.7159  
Age.Group≥70 -0.8105    0.4446   0.4420 -1.834   0.0667 .
ECOG1         0.5506    1.7343   0.5794  0.950   0.3419  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.4715     2.1211    0.1381     1.610
GenderMale      0.8595     1.1634    0.3803     1.943
Age.Group≥70    0.4446     2.2490    0.1870     1.057
ECOG1           1.7343     0.5766    0.5571     5.399

Concordance= 0.629  (se = 0.06 )
Likelihood ratio test= 4.98  on 4 df,   p=0.3
Wald test            = 4.66  on 4 df,   p=0.3
Score (logrank) test = 4.79  on 4 df,   p=0.3

#DFS by ACT treatment in MRD positive - High Risk Stage II

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$Risk.StageII==TRUE,]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

   1588 observations deleted due to missingness 
           n events median 0.95LCL 0.95UCL
ACT=FALSE 15     14   3.52    3.39      NA
ACT=TRUE  23     10     NA    9.30      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Positive ACT vs Observation | High Risk Stage II", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

1588 observations deleted due to missingness 
                ACT=FALSE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
    24.00000      1.00000     13.00000      0.10000      0.08756      0.00781      0.33528 

                ACT=TRUE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000        5.000       10.000        0.537        0.110        0.305        0.722 

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 38, number of events= 24 
   (1588 observations deleted due to missingness)

           coef exp(coef) se(coef)     z Pr(>|z|)    
ACTFALSE 1.6902    5.4206   0.4305 3.926 8.64e-05 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     5.421     0.1845     2.331      12.6

Concordance= 0.709  (se = 0.039 )
Likelihood ratio test= 15.26  on 1 df,   p=9e-05
Wald test            = 15.41  on 1 df,   p=9e-05
Score (logrank) test = 18.65  on 1 df,   p=2e-05

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 5.42 (2.33-12.6); p = 0"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$Risk.StageII==TRUE,]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 38, number of events= 24 
   (1588 observations deleted due to missingness)

                coef exp(coef) se(coef)      z Pr(>|z|)    
ACTFALSE      2.0475    7.7483   0.5070  4.039 5.38e-05 ***
GenderMale   -0.4489    0.6383   0.4772 -0.941   0.3469    
Age.Group≥70  0.2109    1.2348   0.4996  0.422   0.6729    
ECOG1         1.4282    4.1714   0.5908  2.417   0.0156 *  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        7.7483     0.1291    2.8686    20.928
GenderMale      0.6383     1.5666    0.2505     1.626
Age.Group≥70    1.2348     0.8098    0.4638     3.288
ECOG1           4.1714     0.2397    1.3103    13.280

Concordance= 0.759  (se = 0.052 )
Likelihood ratio test= 22.42  on 4 df,   p=2e-04
Wald test            = 19.09  on 4 df,   p=8e-04
Score (logrank) test = 25.15  on 4 df,   p=5e-05

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$Risk.StageII==TRUE,]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 38, number of events= 24 
   (1588 observations deleted due to missingness)

                coef exp(coef) se(coef)      z Pr(>|z|)    
ACTTRUE      -2.0475    0.1291   0.5070 -4.039 5.38e-05 ***
GenderMale   -0.4489    0.6383   0.4772 -0.941   0.3469    
Age.Group≥70  0.2109    1.2348   0.4996  0.422   0.6729    
ECOG1         1.4282    4.1714   0.5908  2.417   0.0156 *  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.1291     7.7483   0.04778    0.3486
GenderMale      0.6383     1.5666   0.25054    1.6264
Age.Group≥70    1.2348     0.8098   0.46377    3.2878
ECOG1           4.1714     0.2397   1.31029   13.2798

Concordance= 0.759  (se = 0.052 )
Likelihood ratio test= 22.42  on 4 df,   p=2e-04
Wald test            = 19.09  on 4 df,   p=8e-04
Score (logrank) test = 25.15  on 4 df,   p=5e-05

#DFS by ACT treatment in MRD negative - Stage II T3N0

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$StageII.Group=="T3N0",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

            n events median 0.95LCL 0.95UCL
ACT=FALSE 400     17     NA      NA      NA
ACT=TRUE   76      1     NA      NA      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Negative ACT vs Observation | T3N0", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     166.0000      16.0000       0.9516       0.0121       0.9212       0.9704 

                ACT=TRUE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      31.0000       1.0000       0.9811       0.0187       0.8735       0.9973 

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 476, number of events= 18 

          coef exp(coef) se(coef)     z Pr(>|z|)
ACTFALSE 1.195     3.304    1.029 1.161    0.246

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     3.304     0.3027    0.4396     24.83

Concordance= 0.559  (se = 0.023 )
Likelihood ratio test= 1.94  on 1 df,   p=0.2
Wald test            = 1.35  on 1 df,   p=0.2
Score (logrank) test = 1.52  on 1 df,   p=0.2

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 3.3 (0.44-24.83); p = 0.246"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$StageII.Group=="T3N0",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 476, number of events= 18 

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTFALSE      1.3971    4.0433   1.0319  1.354   0.1758  
GenderMale    0.1738    1.1898   0.4719  0.368   0.7127  
Age.Group≥70 -1.3071    0.2706   0.5576 -2.344   0.0191 *
ECOG1         0.4088    1.5051   0.7931  0.516   0.6062  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        4.0433     0.2473   0.53501   30.5566
GenderMale      1.1898     0.8405   0.47186    3.0000
Age.Group≥70    0.2706     3.6955   0.09072    0.8072
ECOG1           1.5051     0.6644   0.31805    7.1221

Concordance= 0.688  (se = 0.041 )
Likelihood ratio test= 8.51  on 4 df,   p=0.07
Wald test            = 7.13  on 4 df,   p=0.1
Score (logrank) test = 7.92  on 4 df,   p=0.09

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$StageII.Group=="T3N0",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))

circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 476, number of events= 18 

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTTRUE      -1.3971    0.2473   1.0319 -1.354   0.1758  
GenderMale    0.1738    1.1898   0.4719  0.368   0.7127  
Age.Group≥70 -1.3071    0.2706   0.5576 -2.344   0.0191 *
ECOG1         0.4088    1.5051   0.7931  0.516   0.6062  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.2473     4.0433   0.03273    1.8691
GenderMale      1.1898     0.8405   0.47186    3.0000
Age.Group≥70    0.2706     3.6955   0.09072    0.8072
ECOG1           1.5051     0.6644   0.31805    7.1221

Concordance= 0.688  (se = 0.041 )
Likelihood ratio test= 8.51  on 4 df,   p=0.07
Wald test            = 7.13  on 4 df,   p=0.1
Score (logrank) test = 7.92  on 4 df,   p=0.09

#DFS by ACT treatment in MRD negative - Stage II T4N0

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$StageII.Group=="T4N0",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

           n events median 0.95LCL 0.95UCL
ACT=FALSE 64      9     NA      NA      NA
ACT=TRUE  29      2     NA      NA      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Negative ACT vs Observation | T4N0", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      19.0000       9.0000       0.8478       0.0471       0.7267       0.9181 

                ACT=TRUE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       6.0000       2.0000       0.9205       0.0544       0.7154       0.9797 

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 93, number of events= 11 

           coef exp(coef) se(coef)    z Pr(>|z|)
ACTFALSE 0.6570    1.9290   0.7824 0.84    0.401

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     1.929     0.5184    0.4162      8.94

Concordance= 0.561  (se = 0.06 )
Likelihood ratio test= 0.8  on 1 df,   p=0.4
Wald test            = 0.71  on 1 df,   p=0.4
Score (logrank) test = 0.73  on 1 df,   p=0.4

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 1.93 (0.42-8.94); p = 0.401"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$StageII.Group=="T4N0",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 93, number of events= 11 

                coef exp(coef) se(coef)      z Pr(>|z|)
ACTFALSE      0.6626    1.9399   0.8154  0.813    0.416
GenderMale   -0.1393    0.8700   0.6220 -0.224    0.823
Age.Group≥70 -0.3472    0.7066   0.6563 -0.529    0.597
ECOG1         0.3212    1.3788   0.8365  0.384    0.701

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        1.9399     0.5155    0.3924     9.591
GenderMale      0.8700     1.1494    0.2571     2.944
Age.Group≥70    0.7066     1.4152    0.1952     2.557
ECOG1           1.3788     0.7253    0.2676     7.104

Concordance= 0.588  (se = 0.075 )
Likelihood ratio test= 1.17  on 4 df,   p=0.9
Wald test            = 1.08  on 4 df,   p=0.9
Score (logrank) test = 1.11  on 4 df,   p=0.9

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$StageII.Group=="T4N0",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))

circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 93, number of events= 11 

                coef exp(coef) se(coef)      z Pr(>|z|)
ACTTRUE      -0.6626    0.5155   0.8154 -0.813    0.416
GenderMale   -0.1393    0.8700   0.6220 -0.224    0.823
Age.Group≥70 -0.3472    0.7066   0.6563 -0.529    0.597
ECOG1         0.3212    1.3788   0.8365  0.384    0.701

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.5155     1.9399    0.1043     2.549
GenderMale      0.8700     1.1494    0.2571     2.944
Age.Group≥70    0.7066     1.4152    0.1952     2.557
ECOG1           1.3788     0.7253    0.2676     7.104

Concordance= 0.588  (se = 0.075 )
Likelihood ratio test= 1.17  on 4 df,   p=0.9
Wald test            = 1.08  on 4 df,   p=0.9
Score (logrank) test = 1.11  on 4 df,   p=0.9

#DFS by ACT treatment in MRD negative - Stage III

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "IV")),]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

            n events median 0.95LCL 0.95UCL
ACT=FALSE 213     29     NA      NA      NA
ACT=TRUE  469     52     NA      NA      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Negative ACT vs Observation | Stage III", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(18, 24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   18    115      27    0.848  0.0274        0.785        0.894
   24     63       2    0.829  0.0300        0.760        0.879

                ACT=TRUE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   18    293      42    0.898  0.0150        0.864        0.924
   24    178       6    0.876  0.0173        0.837        0.906

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 682, number of events= 81 

           coef exp(coef) se(coef)     z Pr(>|z|)
ACTFALSE 0.2863    1.3315   0.2319 1.235    0.217

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     1.332      0.751    0.8452     2.098

Concordance= 0.537  (se = 0.028 )
Likelihood ratio test= 1.48  on 1 df,   p=0.2
Wald test            = 1.52  on 1 df,   p=0.2
Score (logrank) test = 1.53  on 1 df,   p=0.2

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 1.33 (0.85-2.1); p = 0.217"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "IV")),]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 682, number of events= 81 

                coef exp(coef) se(coef)      z Pr(>|z|)
ACTFALSE      0.3004    1.3505   0.2340  1.284    0.199
GenderMale    0.2382    1.2690   0.2244  1.062    0.288
Age.Group≥70 -0.1732    0.8410   0.2327 -0.744    0.457
ECOG1         0.1347    1.1442   0.3823  0.352    0.725

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE         1.350     0.7405    0.8537     2.136
GenderMale       1.269     0.7880    0.8175     1.970
Age.Group≥70     0.841     1.1891    0.5329     1.327
ECOG1            1.144     0.8740    0.5409     2.420

Concordance= 0.553  (se = 0.033 )
Likelihood ratio test= 3.24  on 4 df,   p=0.5
Wald test            = 3.28  on 4 df,   p=0.5
Score (logrank) test = 3.29  on 4 df,   p=0.5

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "IV")),]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 682, number of events= 81 

                coef exp(coef) se(coef)      z Pr(>|z|)
ACTTRUE      -0.3004    0.7405   0.2340 -1.284    0.199
GenderMale    0.2382    1.2690   0.2244  1.062    0.288
Age.Group≥70 -0.1732    0.8410   0.2327 -0.744    0.457
ECOG1         0.1347    1.1442   0.3823  0.352    0.725

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.7405      1.350    0.4681     1.171
GenderMale      1.2690      0.788    0.8175     1.970
Age.Group≥70    0.8410      1.189    0.5329     1.327
ECOG1           1.1442      0.874    0.5409     2.420

Concordance= 0.553  (se = 0.033 )
Likelihood ratio test= 3.24  on 4 df,   p=0.5
Wald test            = 3.28  on 4 df,   p=0.5
Score (logrank) test = 3.29  on 4 df,   p=0.5

#DFS by ACT treatment in MRD positive - Stage III

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "IV")),]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

            n events median 0.95LCL 0.95UCL
ACT=FALSE  32     31   3.58    2.57    4.01
ACT=TRUE  122     78  11.27    9.10   16.07

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Positive ACT vs Observation | Stage III", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(18, 24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
    18.00000      1.00000     30.00000      0.03906      0.03744      0.00306      0.16257 

                ACT=TRUE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   18     39      72    0.393  0.0455        0.304        0.481
   24     20       5    0.330  0.0464        0.241        0.421

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 154, number of events= 109 

           coef exp(coef) se(coef)     z Pr(>|z|)    
ACTFALSE 1.4135    4.1105   0.2203 6.417 1.39e-10 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE      4.11     0.2433     2.669      6.33

Concordance= 0.619  (se = 0.021 )
Likelihood ratio test= 33.3  on 1 df,   p=8e-09
Wald test            = 41.18  on 1 df,   p=1e-10
Score (logrank) test = 47.85  on 1 df,   p=5e-12

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 4.11 (2.67-6.33); p = 0"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "IV")),]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 154, number of events= 109 

                 coef exp(coef) se(coef)      z Pr(>|z|)    
ACTFALSE      1.48178   4.40077  0.24173  6.130 8.79e-10 ***
GenderMale    0.02384   1.02413  0.19953  0.119    0.905    
Age.Group≥70 -0.01673   0.98341  0.20368 -0.082    0.935    
ECOG1        -0.20242   0.81675  0.33241 -0.609    0.543    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        4.4008     0.2272    2.7401     7.068
GenderMale      1.0241     0.9764    0.6926     1.514
Age.Group≥70    0.9834     1.0169    0.6597     1.466
ECOG1           0.8168     1.2244    0.4257     1.567

Concordance= 0.63  (se = 0.027 )
Likelihood ratio test= 33.75  on 4 df,   p=8e-07
Wald test            = 41.58  on 4 df,   p=2e-08
Score (logrank) test = 48.35  on 4 df,   p=8e-10

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "IV")),]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 154, number of events= 109 

                 coef exp(coef) se(coef)      z Pr(>|z|)    
ACTTRUE      -1.48178   0.22723  0.24173 -6.130 8.79e-10 ***
GenderMale    0.02384   1.02413  0.19953  0.119    0.905    
Age.Group≥70 -0.01673   0.98341  0.20368 -0.082    0.935    
ECOG1        -0.20242   0.81675  0.33241 -0.609    0.543    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.2272     4.4008    0.1415    0.3649
GenderMale      1.0241     0.9764    0.6926    1.5142
Age.Group≥70    0.9834     1.0169    0.6597    1.4659
ECOG1           0.8168     1.2244    0.4257    1.5669

Concordance= 0.63  (se = 0.027 )
Likelihood ratio test= 33.75  on 4 df,   p=8e-07
Wald test            = 41.58  on 4 df,   p=2e-08
Score (logrank) test = 48.35  on 4 df,   p=8e-10

#DFS by ACT treatment in MRD negative - Stage IV NAC-treated

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

            n events median 0.95LCL 0.95UCL
ACT=FALSE 113     53   27.9    15.3      NA
ACT=TRUE   30     11     NA    20.1      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Negative ACT vs Observation | Stage IV NAC-treated", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(3, 6, 18, 24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3    103      10    0.912  0.0267        0.842        0.951
    6     83      20    0.735  0.0415        0.643        0.806
   18     48      20    0.535  0.0490        0.435        0.625
   24     29       2    0.504  0.0509        0.400        0.598

                ACT=TRUE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3     30       0    1.000  0.0000        1.000        1.000
    6     25       5    0.833  0.0680        0.645        0.927
   18     16       4    0.689  0.0871        0.484        0.825
   24     11       2    0.589  0.0992        0.373        0.753

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 143, number of events= 64 

           coef exp(coef) se(coef)     z Pr(>|z|)
ACTFALSE 0.3749    1.4549   0.3314 1.131    0.258

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     1.455     0.6873    0.7598     2.786

Concordance= 0.535  (se = 0.024 )
Likelihood ratio test= 1.39  on 1 df,   p=0.2
Wald test            = 1.28  on 1 df,   p=0.3
Score (logrank) test = 1.29  on 1 df,   p=0.3

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 1.45 (0.76-2.79); p = 0.258"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 143, number of events= 64 

                coef exp(coef) se(coef)      z Pr(>|z|)
ACTFALSE      0.3908    1.4781   0.3332  1.173    0.241
GenderMale    0.3629    1.4375   0.2635  1.377    0.168
Age.Group≥70 -0.3175    0.7279   0.2697 -1.178    0.239
ECOG1        -0.5519    0.5759   0.7251 -0.761    0.447

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        1.4781     0.6765    0.7694     2.840
GenderMale      1.4375     0.6956    0.8577     2.409
Age.Group≥70    0.7279     1.3737    0.4291     1.235
ECOG1           0.5759     1.7365    0.1390     2.385

Concordance= 0.574  (se = 0.036 )
Likelihood ratio test= 5.95  on 4 df,   p=0.2
Wald test            = 5.5  on 4 df,   p=0.2
Score (logrank) test = 5.6  on 4 df,   p=0.2

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 143, number of events= 64 

                coef exp(coef) se(coef)      z Pr(>|z|)
ACTTRUE      -0.3908    0.6765   0.3332 -1.173    0.241
GenderMale    0.3629    1.4375   0.2635  1.377    0.168
Age.Group≥70 -0.3175    0.7279   0.2697 -1.178    0.239
ECOG1        -0.5519    0.5759   0.7251 -0.761    0.447

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.6765     1.4781    0.3521     1.300
GenderMale      1.4375     0.6956    0.8577     2.409
Age.Group≥70    0.7279     1.3737    0.4291     1.235
ECOG1           0.5759     1.7365    0.1390     2.385

Concordance= 0.574  (se = 0.036 )
Likelihood ratio test= 5.95  on 4 df,   p=0.2
Wald test            = 5.5  on 4 df,   p=0.2
Score (logrank) test = 5.6  on 4 df,   p=0.2

#DFS by ACT treatment in MRD Negative - Stage IV no NAC-treated

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="FALSE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

           n events median 0.95LCL 0.95UCL
ACT=FALSE 81     30     NA    33.1      NA
ACT=TRUE  50     14     NA      NA      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Negative ACT vs Observation | Stage IV No NAC-treated", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(3, 6, 18, 24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3     75       6    0.926  0.0291        0.843        0.966
    6     73       2    0.901  0.0331        0.812        0.949
   18     44      19    0.655  0.0541        0.538        0.750
   24     25       1    0.636  0.0559        0.516        0.734

                ACT=TRUE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3     49       1    0.980  0.0198        0.866        0.997
    6     48       1    0.960  0.0277        0.849        0.990
   18     29       9    0.765  0.0623        0.615        0.863
   24     16       3    0.658  0.0790        0.479        0.787

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 131, number of events= 44 

           coef exp(coef) se(coef)     z Pr(>|z|)
ACTFALSE 0.3877    1.4736   0.3240 1.197    0.231

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     1.474     0.6786    0.7809     2.781

Concordance= 0.56  (se = 0.035 )
Likelihood ratio test= 1.49  on 1 df,   p=0.2
Wald test            = 1.43  on 1 df,   p=0.2
Score (logrank) test = 1.45  on 1 df,   p=0.2

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 1.47 (0.78-2.78); p = 0.231"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="FALSE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 131, number of events= 44 

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTFALSE      0.2636    1.3016   0.3355  0.786   0.4320  
GenderMale    0.3385    1.4029   0.3255  1.040   0.2983  
Age.Group≥70  0.7056    2.0251   0.3167  2.228   0.0259 *
ECOG1        -1.5549    0.2112   1.0202 -1.524   0.1275  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        1.3016     0.7683    0.6744     2.512
GenderMale      1.4029     0.7128    0.7412     2.655
Age.Group≥70    2.0251     0.4938    1.0886     3.767
ECOG1           0.2112     4.7344    0.0286     1.560

Concordance= 0.64  (se = 0.042 )
Likelihood ratio test= 10.99  on 4 df,   p=0.03
Wald test            = 10.06  on 4 df,   p=0.04
Score (logrank) test = 10.67  on 4 df,   p=0.03

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="FALSE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 131, number of events= 44 

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTTRUE      -0.2636    0.7683   0.3355 -0.786   0.4320  
GenderMale    0.3385    1.4029   0.3255  1.040   0.2983  
Age.Group≥70  0.7056    2.0251   0.3167  2.228   0.0259 *
ECOG1        -1.5549    0.2112   1.0202 -1.524   0.1275  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.7683     1.3016    0.3981     1.483
GenderMale      1.4029     0.7128    0.7412     2.655
Age.Group≥70    2.0251     0.4938    1.0886     3.767
ECOG1           0.2112     4.7344    0.0286     1.560

Concordance= 0.64  (se = 0.042 )
Likelihood ratio test= 10.99  on 4 df,   p=0.03
Wald test            = 10.06  on 4 df,   p=0.04
Score (logrank) test = 10.67  on 4 df,   p=0.03

#DFS by ACT treatment in MRD positive - Stage IV NAC-treated

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

           n events median 0.95LCL 0.95UCL
ACT=FALSE 32     32   1.46    0.86    2.44
ACT=TRUE  14     13   3.78    3.13   12.59

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Positive ACT vs Observation | Stage IV NAC-treated", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(3, 6, 18, 24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3      7      25   0.2188  0.0731      0.09649        0.372
    6      2       5   0.0625  0.0428      0.01112        0.181
   18      1       1   0.0312  0.0308      0.00237        0.137
   24      1       0   0.0312  0.0308      0.00237        0.137

                ACT=TRUE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3     10       4   0.7143  0.1207      0.40630        0.882
    6      4       6   0.2857  0.1207      0.08834        0.524
   18      1       3   0.0714  0.0688      0.00452        0.275
   24      1       0   0.0714  0.0688      0.00452        0.275

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 46, number of events= 45 

           coef exp(coef) se(coef)     z Pr(>|z|)  
ACTFALSE 0.7342    2.0839   0.3380 2.172   0.0298 *
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     2.084     0.4799     1.074     4.042

Concordance= 0.591  (se = 0.043 )
Likelihood ratio test= 5.1  on 1 df,   p=0.02
Wald test            = 4.72  on 1 df,   p=0.03
Score (logrank) test = 4.9  on 1 df,   p=0.03

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 2.08 (1.07-4.04); p = 0.03"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 46, number of events= 45 

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTFALSE      0.9147    2.4961   0.3739  2.447   0.0144 *
GenderMale   -0.4952    0.6095   0.3597 -1.377   0.1686  
Age.Group≥70  0.1691    1.1843   0.3357  0.504   0.6145  
ECOG1             NA        NA   0.0000     NA       NA  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        2.4961     0.4006    1.1996     5.194
GenderMale      0.6095     1.6408    0.3011     1.233
Age.Group≥70    1.1843     0.8444    0.6133     2.287
ECOG1               NA         NA        NA        NA

Concordance= 0.637  (se = 0.047 )
Likelihood ratio test= 7  on 3 df,   p=0.07
Wald test            = 6.36  on 3 df,   p=0.1
Score (logrank) test = 6.56  on 3 df,   p=0.09

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 46, number of events= 45 

                coef exp(coef) se(coef)      z Pr(>|z|)  
ACTTRUE      -0.9147    0.4006   0.3739 -2.447   0.0144 *
GenderMale   -0.4952    0.6095   0.3597 -1.377   0.1686  
Age.Group≥70  0.1691    1.1843   0.3357  0.504   0.6145  
ECOG1             NA        NA   0.0000     NA       NA  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.4006     2.4961    0.1925    0.8336
GenderMale      0.6095     1.6408    0.3011    1.2335
Age.Group≥70    1.1843     0.8444    0.6133    2.2868
ECOG1               NA         NA        NA        NA

Concordance= 0.637  (se = 0.047 )
Likelihood ratio test= 7  on 3 df,   p=0.07
Wald test            = 6.36  on 3 df,   p=0.1
Score (logrank) test = 6.56  on 3 df,   p=0.09

#DFS by ACT treatment in MRD positive - Stage IV no NAC-treated

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="FALSE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ACT, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ACT, data = circ_data)

           n events median 0.95LCL 0.95UCL
ACT=FALSE 28     27    2.8    1.12    3.52
ACT=TRUE  26     15   14.2    5.92      NA

event_summary <- circ_data %>%
  group_by(ACT) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ACT, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue"), title="DFS - ctDNA MRD Positive ACT vs Observation | Stage IV No NAC-treated", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Observation", "ACT"), legend.title="")

summary(KM_curve, times= c(3, 6, 18, 24))

Call: survfit(formula = surv_object ~ ACT, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                ACT=FALSE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3     14      14    0.500  0.0945        0.306        0.666
    6      4      10    0.143  0.0661        0.045        0.295

                ACT=TRUE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    3     24       2    0.923  0.0523        0.726         0.98
    6     16       7    0.650  0.0944        0.434         0.80
   18      7       6    0.367  0.1031        0.176         0.56
   24      5       0    0.367  0.1031        0.176         0.56

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
cox_fit <- coxph(surv_object ~ ACT, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT, data = circ_data)

  n= 54, number of events= 42 

           coef exp(coef) se(coef)     z Pr(>|z|)    
ACTFALSE 1.9565    7.0742   0.3975 4.922 8.55e-07 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

         exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE     7.074     0.1414     3.246     15.42

Concordance= 0.714  (se = 0.025 )
Likelihood ratio test= 28.96  on 1 df,   p=7e-08
Wald test            = 24.23  on 1 df,   p=9e-07
Score (logrank) test = 31.08  on 1 df,   p=2e-08

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 7.07 (3.25-15.42); p = 0"

#Adjusted HR "ACT vs no ACT" - age, gender, MSI, pathological stage, and performance status
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="FALSE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("TRUE","FALSE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 54, number of events= 42 

                  coef exp(coef)  se(coef)      z Pr(>|z|)    
ACTFALSE      1.998062  7.374747  0.405506  4.927 8.34e-07 ***
GenderMale   -0.173869  0.840407  0.330194 -0.527    0.598    
Age.Group≥70  0.001981  1.001983  0.319386  0.006    0.995    
ECOG1        -0.038826  0.961919  0.615096 -0.063    0.950    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTFALSE        7.3747     0.1356    3.3310    16.327
GenderMale      0.8404     1.1899    0.4400     1.605
Age.Group≥70    1.0020     0.9980    0.5358     1.874
ECOG1           0.9619     1.0396    0.2881     3.212

Concordance= 0.728  (se = 0.032 )
Likelihood ratio test= 29.24  on 4 df,   p=7e-06
Wald test            = 24.52  on 4 df,   p=6e-05
Score (logrank) test = 31.36  on 4 df,   p=3e-06

#Same analysis; Non ACT as reference
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!(circ_data$Stage %in% c("I", "II", "III")),]
circ_data <- circ_data[circ_data$NAC=="FALSE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data$DFS.months=circ_data$DFS.months-2
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ACT <- factor(circ_data$ACT, levels=c("FALSE","TRUE"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", "≥70"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Stage <- factor(circ_data$Stage, levels = c("II", "III"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$Colon <- factor(circ_data$PrimSite, levels = c("Right-sided colon", "Left-sided colon", "Rectum"))
circ_data$ECOG <- factor(circ_data$ECOG, levels=c("0","1"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-HIGH"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ACT + Gender + Age.Group + ECOG, data=circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ACT + Gender + Age.Group + ECOG, 
    data = circ_data)

  n= 54, number of events= 42 

                  coef exp(coef)  se(coef)      z Pr(>|z|)    
ACTTRUE      -1.998062  0.135598  0.405506 -4.927 8.34e-07 ***
GenderMale   -0.173869  0.840407  0.330194 -0.527    0.598    
Age.Group≥70  0.001981  1.001983  0.319386  0.006    0.995    
ECOG1        -0.038826  0.961919  0.615096 -0.063    0.950    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

             exp(coef) exp(-coef) lower .95 upper .95
ACTTRUE         0.1356      7.375   0.06125    0.3002
GenderMale      0.8404      1.190   0.43998    1.6053
Age.Group≥70    1.0020      0.998   0.53579    1.8738
ECOG1           0.9619      1.040   0.28812    3.2115

Concordance= 0.728  (se = 0.032 )
Likelihood ratio test= 29.24  on 4 df,   p=7e-06
Wald test            = 24.52  on 4 df,   p=6e-05
Score (logrank) test = 31.36  on 4 df,   p=3e-06

#DFS by ctDNA Clearance ACT-treated at 3 months - all stages

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ACT==TRUE,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Dynamics <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Dynamics = case_when(
    ctDNA.MRD == "POSITIVE" & ctDNA.3months == "NEGATIVE" ~ 1,
    ctDNA.MRD == "POSITIVE" & ctDNA.3months == "POSITIVE" ~ 2
  ))

circ_data <- circ_data[circ_data$DFS.3mo.months>=0,]
survfit(Surv(time = circ_data$DFS.3mo.months, event = circ_data$DFS.Event)~ctDNA.Dynamics, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.3mo.months, event = circ_data$DFS.Event) ~ 
    ctDNA.Dynamics, data = circ_data)

   674 observations deleted due to missingness 
                   n events median 0.95LCL 0.95UCL
ctDNA.Dynamics=1 100     42  27.53   18.07      NA
ctDNA.Dynamics=2  71     64   4.14    3.22    5.55

event_summary <- circ_data %>%
  group_by(ctDNA.Dynamics) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.3mo.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Dynamics, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA Clearance from MRD to 3 months ACT-treated | All Stages", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Clearance", "No Clearance"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.Dynamics, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

674 observations deleted due to missingness 
                ctDNA.Dynamics=1 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      23.0000      41.0000       0.5217       0.0571       0.4047       0.6263 

                ctDNA.Dynamics=2 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       5.0000      64.0000       0.0913       0.0355       0.0372       0.1753 

circ_data$ctDNA.Dynamics <- factor(circ_data$ctDNA.Dynamics, levels=c("1","2"), labels = c("Clearance", "No Clearance"))
cox_fit <- coxph(surv_object ~ ctDNA.Dynamics, data=circ_data) 
ggforest(cox_fit,data = circ_data) 

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Dynamics, data = circ_data)

  n= 171, number of events= 106 
   (674 observations deleted due to missingness)

                             coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.DynamicsNo Clearance 1.6822    5.3775   0.2055 8.187 2.67e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.DynamicsNo Clearance     5.378      0.186     3.595     8.044

Concordance= 0.716  (se = 0.018 )
Likelihood ratio test= 67.62  on 1 df,   p=<2e-16
Wald test            = 67.03  on 1 df,   p=3e-16
Score (logrank) test = 81.18  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 5.38 (3.59-8.04); p = 0"

#OS by ctDNA Clearance ACT-treated at 3 months - all stages

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ACT==TRUE,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Dynamics <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Dynamics = case_when(
    ctDNA.MRD == "POSITIVE" & ctDNA.3months == "NEGATIVE" ~ 1,
    ctDNA.MRD == "POSITIVE" & ctDNA.3months == "POSITIVE" ~ 2
  ))

circ_data <- circ_data[circ_data$OS.3mo.months>=0,]
survfit(Surv(time = circ_data$OS.3mo.months, event = circ_data$OS.Event)~ctDNA.Dynamics, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.3mo.months, event = circ_data$OS.Event) ~ 
    ctDNA.Dynamics, data = circ_data)

   674 observations deleted due to missingness 
                   n events median 0.95LCL 0.95UCL
ctDNA.Dynamics=1 100      7     NA      NA      NA
ctDNA.Dynamics=2  71     16   41.6    31.9      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Dynamics) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$OS.3mo.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Dynamics, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="OS - ctDNA Clearance from MRD to 3 months ACT-treated | All Stages", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Clearance", "No Clearance"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.Dynamics, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

674 observations deleted due to missingness 
                ctDNA.Dynamics=1 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      39.0000       6.0000       0.8936       0.0423       0.7738       0.9519 

                ctDNA.Dynamics=2 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000       24.000       13.000        0.716        0.070        0.553        0.829 

circ_data$ctDNA.Dynamics <- factor(circ_data$ctDNA.Dynamics, levels=c("1","2"), labels = c("Clearance", "No Clearance"))
cox_fit <- coxph(surv_object ~ ctDNA.Dynamics, data=circ_data) 
ggforest(cox_fit,data = circ_data) 

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Dynamics, data = circ_data)

  n= 171, number of events= 23 
   (674 observations deleted due to missingness)

                             coef exp(coef) se(coef)     z Pr(>|z|)   
ctDNA.DynamicsNo Clearance 1.3251    3.7627   0.4583 2.892  0.00383 **
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.DynamicsNo Clearance     3.763     0.2658     1.533     9.238

Concordance= 0.689  (se = 0.047 )
Likelihood ratio test= 9.18  on 1 df,   p=0.002
Wald test            = 8.36  on 1 df,   p=0.004
Score (logrank) test = 9.65  on 1 df,   p=0.002

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 3.76 (1.53-9.24); p = 0.004"

#DFS by ctDNA Clearance ACT-treated at 6 months - all stages

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ACT==TRUE,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Dynamics <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Dynamics = case_when(
    ctDNA.MRD == "POSITIVE" & ctDNA.6months == "NEGATIVE" ~ 1,
    ctDNA.MRD == "POSITIVE" & ctDNA.6months == "POSITIVE" ~ 2
  ))

circ_data <- circ_data[circ_data$DFS.6mo.months>=0,]
survfit(Surv(time = circ_data$DFS.6mo.months, event = circ_data$DFS.Event)~ctDNA.Dynamics, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.6mo.months, event = circ_data$DFS.Event) ~ 
    ctDNA.Dynamics, data = circ_data)

   732 observations deleted due to missingness 
                  n events median 0.95LCL 0.95UCL
ctDNA.Dynamics=1 77     27     NA   17.74      NA
ctDNA.Dynamics=2 35     34    2.4    1.61    3.68

event_summary <- circ_data %>%
  group_by(ctDNA.Dynamics) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.6mo.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Dynamics, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA Clearance from MRD to 6 months ACT-treated | All Stages", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Clearance", "No Clearance"), legend.title="")

summary(KM_curve, times= c(6, 24))

Call: survfit(formula = surv_object ~ ctDNA.Dynamics, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

732 observations deleted due to missingness 
                ctDNA.Dynamics=1 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    6     61      14    0.816  0.0445        0.709        0.886
   24     15      13    0.602  0.0625        0.469        0.712

                ctDNA.Dynamics=2 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      6.0000       5.0000      29.0000       0.1607       0.0638       0.0609       0.3028 

circ_data$ctDNA.Dynamics <- factor(circ_data$ctDNA.Dynamics, levels=c("1","2"), labels = c("Clearance", "No Clearance"))
cox_fit <- coxph(surv_object ~ ctDNA.Dynamics, data=circ_data) 
ggforest(cox_fit,data = circ_data) 

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Dynamics, data = circ_data)

  n= 112, number of events= 61 
   (732 observations deleted due to missingness)

                              coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.DynamicsNo Clearance  2.4088   11.1201   0.3069 7.848 4.24e-15 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.DynamicsNo Clearance     11.12    0.08993     6.093     20.29

Concordance= 0.729  (se = 0.023 )
Likelihood ratio test= 64.06  on 1 df,   p=1e-15
Wald test            = 61.58  on 1 df,   p=4e-15
Score (logrank) test = 88.6  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 11.12 (6.09-20.29); p = 0"

#OS by ctDNA Clearance ACT-treated at 6 months - all stages

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$ACT==TRUE,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Dynamics <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Dynamics = case_when(
    ctDNA.MRD == "POSITIVE" & ctDNA.6months == "NEGATIVE" ~ 1,
    ctDNA.MRD == "POSITIVE" & ctDNA.6months == "POSITIVE" ~ 2
  ))

circ_data <- circ_data[circ_data$OS.6mo.months>=0,]
survfit(Surv(time = circ_data$OS.6mo.months, event = circ_data$OS.Event)~ctDNA.Dynamics, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.6mo.months, event = circ_data$OS.Event) ~ 
    ctDNA.Dynamics, data = circ_data)

   732 observations deleted due to missingness 
                  n events median 0.95LCL 0.95UCL
ctDNA.Dynamics=1 77      3     NA      NA      NA
ctDNA.Dynamics=2 36      7     39    27.9      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Dynamics) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$OS.6mo.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Dynamics, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="OS - ctDNA Clearance from MRD to 6 months ACT-treated | All Stages", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("Clearance", "No Clearance"), legend.title="")

summary(KM_curve, times= c(6, 24))

Call: survfit(formula = surv_object ~ ctDNA.Dynamics, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

732 observations deleted due to missingness 
                ctDNA.Dynamics=1 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    6     72       0    1.000  0.0000           NA           NA
   24     27       2    0.966  0.0236        0.871        0.991

                ctDNA.Dynamics=2 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    6     22       3    0.896  0.0571        0.710        0.966
   24      8       2    0.791  0.0863        0.558        0.910

circ_data$ctDNA.Dynamics <- factor(circ_data$ctDNA.Dynamics, levels=c("1","2"), labels = c("Clearance", "No Clearance"))
cox_fit <- coxph(surv_object ~ ctDNA.Dynamics, data=circ_data) 
ggforest(cox_fit,data = circ_data) 

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Dynamics, data = circ_data)

  n= 113, number of events= 10 
   (732 observations deleted due to missingness)

                             coef exp(coef) se(coef)     z Pr(>|z|)   
ctDNA.DynamicsNo Clearance 1.8445    6.3252   0.7088 2.602  0.00926 **
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.DynamicsNo Clearance     6.325     0.1581     1.577     25.37

Concordance= 0.747  (se = 0.071 )
Likelihood ratio test= 7.27  on 1 df,   p=0.007
Wald test            = 6.77  on 1 df,   p=0.009
Score (logrank) test = 8.89  on 1 df,   p=0.003

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 6.33 (1.58-25.37); p = 0.009"

#Number of MRD positive patients & ctDNA clearance on ACT

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

# Count the number of MRD positive patients
number_of_positive_patients <- sum(circ_datadf$ctDNA.MRD == "POSITIVE", na.rm = TRUE)
print(paste("Number of MRD positive patients:", number_of_positive_patients))

[1] "Number of MRD positive patients: 336"

# Count the number & percentage of MRD positive patients treated with ACT
positive_subset <- sum(circ_datadf$ACT == "TRUE" & circ_datadf$ctDNA.MRD == "POSITIVE", na.rm = TRUE)
print(paste("Number of MRD positive patients treated with ACT:", positive_subset))

[1] "Number of MRD positive patients treated with ACT: 185"

percentage_positive_for_both <- (positive_subset / number_of_positive_patients) * 100
print(paste("Percentage of MRD positive patients treated with ACT:", percentage_positive_for_both, "%"))

[1] "Percentage of MRD positive patients treated with ACT: 55.0595238095238 %"

# Count the number & percentage of patients with ctDNA clearance post-ACT
clearance_postACT <- sum(
  (circ_datadf$ACT == "TRUE") & 
    (circ_datadf$ctDNA.MRD == "POSITIVE") & 
    (circ_datadf$Clearance.Event == "TRUE"), 
  na.rm = TRUE
)
print(paste("Number of patients with ctDNA Clearance post-ACT:", clearance_postACT))

[1] "Number of patients with ctDNA Clearance post-ACT: 126"

percentage_clearance <- (clearance_postACT / positive_subset) * 100
print(paste("ctDNA Clearance post-ACT:", percentage_clearance, "%"))

[1] "ctDNA Clearance post-ACT: 68.1081081081081 %"

# Count the number of patients with subsequent timepoints available
clearance_subset <- sum(
  (circ_datadf$ACT == "TRUE") & 
    (circ_datadf$ctDNA.MRD == "POSITIVE") & 
    (circ_datadf$Transient.Clearance == "TRUE" | circ_datadf$Transient.Clearance == "FALSE"), 
  na.rm = TRUE
)
print(paste("Number of patients with subsequent timepoints available:", clearance_subset))

[1] "Number of patients with subsequent timepoints available: 126"

# Count the number & percentage of patients with sustained clearance
clearance_sustained <- sum(
  (circ_datadf$ACT == "TRUE") & 
    (circ_datadf$ctDNA.MRD == "POSITIVE") & 
    (circ_datadf$Transient.Clearance == "FALSE"), 
  na.rm = TRUE
)
print(paste("Number of patients with sustained clearance:", clearance_sustained))

[1] "Number of patients with sustained clearance: 68"

percentage_sustained_clearance <- (clearance_sustained / clearance_subset) * 100
print(paste("Sustained ctDNA Clearance:", percentage_sustained_clearance, "%"))

[1] "Sustained ctDNA Clearance: 53.968253968254 %"

# Count the number & percentage of patients with transient clearance
clearance_transient <- sum(
  (circ_datadf$ACT == "TRUE") & 
    (circ_datadf$ctDNA.MRD == "POSITIVE") & 
    (circ_datadf$Transient.Clearance == "TRUE"), 
  na.rm = TRUE
)
print(paste("Number of patients with transient clearance:", clearance_transient))

[1] "Number of patients with transient clearance: 58"

percentage_transient_clearance <- (clearance_transient / clearance_subset) * 100
print(paste("Transient ctDNA Clearance:", percentage_transient_clearance, "%"))

[1] "Transient ctDNA Clearance: 46.031746031746 %"

#Sankey plot for Sustained vs Transient Clearance

##To run this commands, please visit: https://sankeymatic.com/build/
#ctDNA + MRD window [185] ACT-treated #ADD8E6
#ctDNA + MRD window [151] Not treated #808080
#ACT-treated [126] ctDNA post-MRD Clearance #87EA86
#ACT-treated [55] No Clearance #E67272
#ACT-treated [4] No post-MRD time point #808080
#No Clearance [55] No Clearance analysis #E67272
#ctDNA post-MRD Clearance [126] Available post-MRD Timepoints #ADD8E66
#Available post-MRD Timepoints [68] Sustained Clearance #7393B3
#Available post-MRD Timepoints [58] Transient Clearance #87EA86

#DFS by ctDNA Clearance post-MRD - 3 Groups

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_data <- circ_data[circ_data$ctDNA.Clearance!="",]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.Clearance, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.Clearance, data = circ_data)

   131 observations deleted due to missingness 
                              n events median 0.95LCL 0.95UCL
ctDNA.Clearance=No Clearance 55     55   4.83    4.53    5.45
ctDNA.Clearance=Sustained    68      7     NA      NA      NA
ctDNA.Clearance=Transient    58     50  12.88   10.38   15.64

event_summary <- circ_data %>%
  group_by(ctDNA.Clearance) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Clearance, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue","green"), title="DFS - ctDNA Clearance post-MRD | All Stages", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("No Clearance", "Sustained", "Transient"), legend.title="")

summary(KM_curve, times= c(12, 18, 24))

Call: survfit(formula = surv_object ~ ctDNA.Clearance, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

131 observations deleted due to missingness 
                ctDNA.Clearance=No Clearance 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
    12.00000      1.00000     54.00000      0.01818      0.01802      0.00149      0.08474 

                ctDNA.Clearance=Sustained 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12     57       5    0.925  0.0321        0.830        0.968
   18     48       1    0.909  0.0354        0.809        0.958
   24     31       1    0.890  0.0394        0.783        0.946

                ctDNA.Clearance=Transient 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12     28      27   0.5212  0.0668      0.38358        0.642
   18      6      18   0.1500  0.0527      0.06542        0.267
   24      1       4   0.0333  0.0312      0.00294        0.137

circ_data$ctDNA.Clearance <- factor(circ_data$ctDNA.Clearance, levels=c("Sustained","Transient", "No Clearance"))
cox_fit <- coxph(surv_object ~ ctDNA.Clearance, data=circ_data) 
ggforest(cox_fit,data = circ_data) 

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Clearance, data = circ_data)

  n= 181, number of events= 112 
   (131 observations deleted due to missingness)

                                coef exp(coef) se(coef)      z Pr(>|z|)    
ctDNA.ClearanceTransient      2.9815   19.7182   0.4229  7.051 1.78e-12 ***
ctDNA.ClearanceNo Clearance   4.8264  124.7631   0.4565 10.573  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                            exp(coef) exp(-coef) lower .95 upper .95
ctDNA.ClearanceTransient        19.72   0.050715     8.608     45.17
ctDNA.ClearanceNo Clearance    124.76   0.008015    50.996    305.24

Concordance= 0.83  (se = 0.017 )
Likelihood ratio test= 207  on 2 df,   p=<2e-16
Wald test            = 129.1  on 2 df,   p=<2e-16
Score (logrank) test = 234.2  on 2 df,   p=<2e-16

#Levels of MRD MTM/mL in Clearance post-MRD log10 transformation

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[!is.na(circ_data$ctDNA.Clearance) & circ_data$ctDNA.Clearance != "",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_data <- as.data.frame(circ_data)

# Transform p_MRD_MTM with log10
circ_data$p_MRD_MTM <- as.numeric(as.character(circ_data$p_MRD_MTM))
circ_data$ctDNA.Clearance <- factor(circ_data$ctDNA.Clearance, levels=c("Sustained","Transient", "No Clearance"))
median_p_MRD_MTM <- aggregate(p_MRD_MTM ~ ctDNA.Clearance, data = circ_data, FUN = median)
print(median_p_MRD_MTM)

# Create violin plot with log10 scale on y-axis
ggplot(circ_data, aes(x=ctDNA.Clearance, y=p_MRD_MTM, fill=ctDNA.Clearance)) +
  geom_violin(trim=FALSE) +
  scale_fill_manual(values=c("Sustained"="lightblue", "Transient"="lightgreen", "No Clearance"="salmon")) +
  geom_boxplot(width=0.1, fill="white", colour="black", alpha=0.5) +
  scale_y_log10(breaks=c(0.001, 0.01, 0.1, 1, 10, 100, 1000, 10000)) +
  labs(title="MRD MTM/mL | Clearance post-MRD", x="Clearance post-MRD", y="MRD MTM/mL") +
  theme_minimal() +
  theme(legend.position="none")

m3_1v2 <- wilcox.test(p_MRD_MTM ~ ctDNA.Clearance,
                      data = circ_data[circ_data$ctDNA.Clearance %in% c("Sustained", "Transient"), ],
                      na.rm = TRUE)
print(m3_1v2)

    Wilcoxon rank sum test with continuity correction

data:  p_MRD_MTM by ctDNA.Clearance
W = 1946, p-value = 0.9007
alternative hypothesis: true location shift is not equal to 0

m3_1v3 <- wilcox.test(p_MRD_MTM ~ ctDNA.Clearance,
                      data = circ_data[circ_data$ctDNA.Clearance %in% c("Sustained", "No Clearance"), ],
                      na.rm = TRUE)
print(m3_1v3)

    Wilcoxon rank sum test with continuity correction

data:  p_MRD_MTM by ctDNA.Clearance
W = 906, p-value = 9.529e-07
alternative hypothesis: true location shift is not equal to 0

m3_2v3 <- wilcox.test(p_MRD_MTM ~ ctDNA.Clearance,
                      data = circ_data[circ_data$ctDNA.Clearance %in% c("Transient", "No Clearance"), ],
                      na.rm = TRUE)
print(m3_2v3)

    Wilcoxon rank sum test with continuity correction

data:  p_MRD_MTM by ctDNA.Clearance
W = 782, p-value = 3.052e-06
alternative hypothesis: true location shift is not equal to 0

#Percentages of recurred transient clearance that return positive

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data <- circ_data[circ_data$ACT=="TRUE",]
circ_data <- circ_data[circ_data$Clearance.Event=="TRUE",]
circ_data <- circ_data[circ_data$DFS.Event=="TRUE",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_data <- subset(circ_data, !is.na(Transient.Clearance))
circ_data <- circ_data[circ_data$Transient.Clearance=="TRUE",]
circ_datadf <- as.data.frame(circ_data)

# Convert days to months
circ_data$p_drelReturned_months <- circ_data$p_drelReturned / 30.437

# Define the intervals: 6-9, 9-12, 12-15, 15-18, 18-21, 21-24, >24 months
breaks <- c(3, 6, 9, 12, 15, 18, 21, 24, 27)
labels <- c("3-6m", "6-9m", "9-12m", "12-15m", "15-18m", "18-21m", "21-24m", ">24m")

# Categorize p_drelReturned_months into intervals
circ_data$p_drelReturned_intervals <- cut(circ_data$p_drelReturned_months, breaks = breaks, labels = labels, right = FALSE)

# Examine the distribution of the intervals
table(circ_data$p_drelReturned_intervals)

  3-6m   6-9m  9-12m 12-15m 15-18m 18-21m 21-24m   >24m 
     7     23      8      4      6      0      2      0 

# Get the counts for each interval
interval_counts <- table(circ_data$p_drelReturned_intervals)

# Calculate the percentages
interval_percentages <- 100 * interval_counts / sum(interval_counts)

# Combine the counts and percentages for a clearer overview
interval_summary <- data.frame(Counts = interval_counts, Percentages = interval_percentages)

# Print the summary
print(interval_summary)

# Calculate cumulative percentages
cumulative_percentages <- cumsum(interval_percentages)

# Combine the counts and percentages for a clearer overview
interval_summary <- data.frame(Counts = interval_counts, Percentages = interval_percentages, CumulativePercentages = cumulative_percentages)

bp <- barplot(interval_percentages, 
        main="Distribution of ctDNA Intervals", 
        xlab="Intervals", 
        ylab="Percentage", 
        col="lightblue",
        ylim=c(0, 100),
        las=2) # las=2 makes the axis labels perpendicular to the axis


# Add the cumulative percentages to the plot
points(bp, cumulative_percentages, type="o", pch=22, col="red", cex=1.5)

#OS by ctDNA Clearance post-MRD - 3 Groups

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$Clearance.Cohort=="TRUE",]
circ_datadf <- as.data.frame(circ_data)
surv_object <- Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)

survfit(Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)~ctDNA.Clearance, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event) ~ 
    ctDNA.Clearance, data = circ_data)

                              n events median 0.95LCL 0.95UCL
ctDNA.Clearance=No Clearance 55     17   32.5    23.9      NA
ctDNA.Clearance=Sustained    68      0     NA      NA      NA
ctDNA.Clearance=Transient    58      7     NA      NA      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Clearance) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

KM_curve <- survfit(surv_object ~ ctDNA.Clearance, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("red","blue","green"), title="OS - ctDNA Clearance post-MRD | All Stages", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("No Clearance", "Sustained", "Transient"), legend.title="")

summary(KM_curve, times= c(12, 18, 24))

Call: survfit(formula = surv_object ~ ctDNA.Clearance, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.Clearance=No Clearance 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12     27       7    0.839  0.0570        0.687        0.921
   18     21       4    0.706  0.0776        0.524        0.829
   24     14       2    0.617  0.0895        0.419        0.765

                ctDNA.Clearance=Sustained 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12     61       0        1       0           NA           NA
   18     54       0        1       0           NA           NA
   24     37       0        1       0           NA           NA

                ctDNA.Clearance=Transient 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12     44       0    1.000  0.0000           NA           NA
   18     34       1    0.972  0.0274        0.819        0.996
   24     18       4    0.823  0.0747        0.615        0.925

circ_data$ctDNA.Clearance <- as.factor(circ_data$ctDNA.Clearance)
circ_data$ctDNA.Clearance <- factor(circ_data$ctDNA.Clearance, levels=c("Sustained","Transient", "No Clearance"))
cox_fit <- coxphf(surv_object ~ ctDNA.Clearance, data=circ_data) 
summary(cox_fit)

coxphf(formula = surv_object ~ ctDNA.Clearance, data = circ_data)

Model fitted by Penalized ML
Confidence intervals and p-values by Profile Likelihood 

                                coef se(coef) exp(coef) lower 0.95 upper 0.95    Chisq            p
ctDNA.ClearanceTransient    3.239402 1.510369  25.51846   3.099164   3314.725 11.55743 6.747909e-04
ctDNA.ClearanceNo Clearance 4.325656 1.484378  75.61513  10.218215   9650.929 34.76657 3.717015e-09

Likelihood ratio test=34.78097 on 2 df, p=2.80161e-08, n=181
Wald test = 12.97638 on 2 df, p = 0.001521303

Covariance-Matrix:
                            ctDNA.ClearanceTransient ctDNA.ClearanceNo Clearance
ctDNA.ClearanceTransient                    2.281214                    2.138730
ctDNA.ClearanceNo Clearance                 2.138730                    2.203378

#Percentages of MRD negative with molecular recurrence (returned positive) post-MRD

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD != "" & circ_data$Lead.Time >= 0, ]
circ_data <- circ_data[circ_data$ctDNA.MRD=="NEGATIVE",]
circ_data <- circ_data[circ_data$PostMRDPos.Event=="TRUE",]
circ_datadf <- as.data.frame(circ_data)

# Convert days to months
#circ_data$PostMRDPos.months <- circ_data$PostMRDPos / 30.437

# Define the intervals: 0-6, 6-9, 9-12, 12-15, 15-18, 18-21, 21-24, >24 months
breaks <- c(0, 6, 9, 12, 15, 18, 21, 24, 48)
labels <- c("0-6m", "6-9m", "9-12m", "12-15m", "15-18m", "18-21m", "21-24m", ">24m")

# Categorize p_drelReturned_months into intervals
circ_data$p_drelReturned_intervals <- cut(circ_data$PostMRDPos.months, breaks = breaks, labels = labels, right = FALSE)

# Examine the distribution of the intervals
table(circ_data$p_drelReturned_intervals)

  0-6m   6-9m  9-12m 12-15m 15-18m 18-21m 21-24m   >24m 
    77     35     23      2     20      1      7      0 

# Get the counts for each interval
interval_counts <- table(circ_data$p_drelReturned_intervals)

# Calculate the percentages
interval_percentages <- 100 * interval_counts / sum(interval_counts)

# Combine the counts and percentages for a clearer overview
interval_summary <- data.frame(Counts = interval_counts, Percentages = interval_percentages)

# Calculate the total number of observations
total_observations <- sum(interval_counts)

# Add the total number of observations to the summary
interval_summary$TotalObservations <- c(rep(NA, length(interval_counts)-1), total_observations)

# Print the summary with total observations
print(interval_summary)

# Calculate cumulative percentages
cumulative_percentages <- cumsum(interval_percentages)

# Combine the counts, percentages, and cumulative percentages for a clearer overview
interval_summary <- data.frame(Counts = interval_counts, Percentages = interval_percentages, CumulativePercentages = cumulative_percentages, TotalObservations = c(rep(NA, length(interval_counts)-1), total_observations))

bp <- barplot(interval_percentages, 
              main="Distribution of ctDNA Intervals", 
              xlab="Intervals", 
              ylab="Percentage", 
              col="lightblue",
              ylim=c(0, 100),
              las=2) # las=2 makes the axis labels perpendicular to the axis

# Add the cumulative percentages to the plot
points(bp, cumulative_percentages, type="o", pch=22, col="red", cex=1.5)

print(interval_summary)

#OS by ctDNA MRD positive vs ctDNA negative with molecular recurrence at Surveillance - 3 groups

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Dynamics <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Dynamics 
         = case_when(
    ctDNA.MRD == "NEGATIVE" & ctDNA.Surveillance=="NEGATIVE" ~ 1,
    ctDNA.MRD == "NEGATIVE" & ctDNA.Surveillance=="POSITIVE" ~ 2,
    ctDNA.MRD == "POSITIVE" ~ 3
  ))

circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
survfit(Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)~ctDNA.Dynamics, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event) ~ 
    ctDNA.Dynamics, data = circ_data)

   321 observations deleted due to missingness 
                    n events median 0.95LCL 0.95UCL
ctDNA.Dynamics=1 1294     13     NA      NA      NA
ctDNA.Dynamics=2  159     15     NA      NA      NA
ctDNA.Dynamics=3  336     52   43.4      NA      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Dynamics) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Dynamics, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","green","red"), title="OS - ctDNA MRD Pos vs Neg with Molecular Recurrence at Surveillance Window", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("All-time negative","Molecular Recurrence", "ctDNA MRD Positive"), legend.title="")

summary(KM_curve, times= c(12, 24))

Call: survfit(formula = surv_object ~ ctDNA.Dynamics, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

321 observations deleted due to missingness 
                ctDNA.Dynamics=1 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12   1137       0    1.000  0.0000           NA           NA
   24    640       5    0.995  0.0023        0.988        0.998

                ctDNA.Dynamics=2 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12    126       2    0.987 0.00909        0.949        0.997
   24     58       8    0.900 0.03138        0.817        0.946

                ctDNA.Dynamics=3 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   12    228      17    0.942  0.0136        0.909        0.964
   24    119      20    0.837  0.0258        0.778        0.881

circ_data$ctDNA.Dynamics <- factor(circ_data$ctDNA.Dynamics, levels=c("1","2","3"), labels = c("All-time negative","Molecular Recurrence", "ctDNA MRD Positive"))
cox_fit <- coxph(surv_object ~ ctDNA.Dynamics, data=circ_data) 
ggforest(cox_fit,data = circ_data) 

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Dynamics, data = circ_data)

  n= 1789, number of events= 80 
   (321 observations deleted due to missingness)

                                      coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.DynamicsMolecular Recurrence  2.4747   11.8787   0.3796 6.519 7.09e-11 ***
ctDNA.DynamicsctDNA MRD Positive    3.0205   20.5007   0.3103 9.734  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                                   exp(coef) exp(-coef) lower .95 upper .95
ctDNA.DynamicsMolecular Recurrence     11.88    0.08418     5.644     25.00
ctDNA.DynamicsctDNA MRD Positive       20.50    0.04878    11.160     37.66

Concordance= 0.833  (se = 0.019 )
Likelihood ratio test= 138.3  on 2 df,   p=<2e-16
Wald test            = 94.79  on 2 df,   p=<2e-16
Score (logrank) test = 182.9  on 2 df,   p=<2e-16

rm(list=ls()) #repeat to compare Molecular Recurrence vs ctDNA MRD positive
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Dynamics <- NA #first we create the variable for the ctDNA & NAC combination, and we assign values
circ_data <- circ_data %>%
  mutate(ctDNA.Dynamics = case_when(
    ctDNA.MRD == "NEGATIVE" & ctDNA.Surveillance=="NEGATIVE" ~ 1,
    ctDNA.MRD == "NEGATIVE" & ctDNA.Surveillance=="POSITIVE" ~ 2,
    ctDNA.MRD == "POSITIVE" ~ 3
  ))

circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
surv_object <-Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)
circ_data$ctDNA.Dynamics <- factor(circ_data$ctDNA.Dynamics, levels=c("2","3","1"), labels = c("Molecular Recurrence", "ctDNA MRD Positive", "All-time negative"))
cox_fit <- coxph(surv_object ~ ctDNA.Dynamics, data=circ_data) 
ggforest(cox_fit,data = circ_data) 

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Dynamics, data = circ_data)

  n= 1789, number of events= 80 
   (321 observations deleted due to missingness)

                                     coef exp(coef) se(coef)      z Pr(>|z|)    
ctDNA.DynamicsctDNA MRD Positive  0.54572   1.72584  0.29355  1.859    0.063 .  
ctDNA.DynamicsAll-time negative  -2.47474   0.08418  0.37964 -6.519 7.09e-11 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                                 exp(coef) exp(-coef) lower .95 upper .95
ctDNA.DynamicsctDNA MRD Positive   1.72584     0.5794    0.9708    3.0681
ctDNA.DynamicsAll-time negative    0.08418    11.8787    0.0400    0.1772

Concordance= 0.833  (se = 0.019 )
Likelihood ratio test= 138.3  on 2 df,   p=<2e-16
Wald test            = 94.79  on 2 df,   p=<2e-16
Score (logrank) test = 182.9  on 2 df,   p=<2e-16

#DFS by ctDNA at the Surveillance Window - All stages Landmark 10 weeks

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.Surveillance!="",]
circ_data$DFS.months=circ_data$DFS.months-2.5
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)~ctDNA.Surveillance, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) ~ 
    ctDNA.Surveillance, data = circ_data)

                               n events median 0.95LCL 0.95UCL
ctDNA.Surveillance=NEGATIVE 1481     89     NA      NA      NA
ctDNA.Surveillance=POSITIVE  310    261   8.47    7.09    8.74

event_summary <- circ_data %>%
  group_by(ctDNA.Surveillance) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Surveillance, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA Surveillance window | All stages", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24, 30, 36))

Call: survfit(formula = surv_object ~ ctDNA.Surveillance, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.Surveillance=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24    565      81    0.932 0.00756        0.915        0.945
   30    311       5    0.922 0.00878        0.902        0.937
   36    113       2    0.915 0.00975        0.894        0.933

                ctDNA.Surveillance=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24     14     257   0.0893  0.0197       0.0556        0.133
   30      4       2   0.0649  0.0213       0.0314        0.115

circ_data$ctDNA.Surveillance <- factor(circ_data$ctDNA.Surveillance, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.Surveillance, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Surveillance, data = circ_data)

  n= 1791, number of events= 350 

                              coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.SurveillancePOSITIVE  3.5133   33.5603   0.1289 27.26   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.SurveillancePOSITIVE     33.56     0.0298     26.07      43.2

Concordance= 0.835  (se = 0.01 )
Likelihood ratio test= 875  on 1 df,   p=<2e-16
Wald test            = 743.2  on 1 df,   p=<2e-16
Score (logrank) test = 1682  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 33.56 (26.07-43.2); p = 0"

#OS by ctDNA at the Surveillance Window - All stages Landmark 10 weeks

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.Surveillance!="",]
circ_data$OS.months=circ_data$OS.months-2.5
circ_data <- circ_data[circ_data$OS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$OS.months, event = circ_data$OS.Event)~ctDNA.Surveillance, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.months, event = circ_data$OS.Event) ~ 
    ctDNA.Surveillance, data = circ_data)

                               n events median 0.95LCL 0.95UCL
ctDNA.Surveillance=NEGATIVE 1481     13     NA      NA      NA
ctDNA.Surveillance=POSITIVE  313     41   41.8    37.3      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Surveillance) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$OS.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Surveillance, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="OS - ctDNA Surveillance window | All stages", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24, 30, 36))

Call: survfit(formula = surv_object ~ ctDNA.Surveillance, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.Surveillance=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24    686       7    0.993 0.00288        0.984        0.997
   30    384       5    0.982 0.00552        0.967        0.990
   36    123       1    0.979 0.00608        0.963        0.989

                ctDNA.Surveillance=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24    102      31    0.832  0.0294        0.765        0.881
   30     60       4    0.792  0.0343        0.715        0.850
   36     14       4    0.705  0.0571        0.577        0.801

circ_data$ctDNA.Surveillance <- factor(circ_data$ctDNA.Surveillance, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.Surveillance, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Surveillance, data = circ_data)

  n= 1794, number of events= 54 

                              coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.SurveillancePOSITIVE  2.9708   19.5075   0.3189 9.317   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.SurveillancePOSITIVE     19.51    0.05126     10.44     36.44

Concordance= 0.825  (se = 0.028 )
Likelihood ratio test= 105.6  on 1 df,   p=<2e-16
Wald test            = 86.8  on 1 df,   p=<2e-16
Score (logrank) test = 171.6  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 19.51 (10.44-36.44); p = 0"

#Multivariate cox regression at Surveillance Window for DFS - All stages Landmark 10 weeks

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.Surveillance!="",]
circ_data$DFS.months=circ_data$DFS.months-2.5
circ_data <- circ_data[circ_data$DFS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

circ_data$ctDNA.Surveillance <- factor(circ_data$ctDNA.Surveillance, levels=c("NEGATIVE","POSITIVE"), labels = c("Negative", "Positive"))
circ_data$Gender <- factor(circ_data$Gender, levels = c("Female", "Male"))
circ_data$Age.Group <- factor(circ_data$Age.Group, levels = c("1", "2"), labels = c("<70", ">70"))
circ_data$PrimSite <- factor(circ_data$PrimSite, levels = c("Left-sided colon", "Right-sided colon"))
circ_data$ECOG <- factor(circ_data$ECOG, levels = c("0", "1"))
circ_data$pT <- factor(circ_data$pT, levels = c("T1-T2", "T3-T4"))
circ_data$pN <- factor(circ_data$pN, levels = c("N0", "N1-N2"))
circ_data$MSI <- factor(circ_data$MSI, levels = c("MSS", "MSI-High"), labels = c("MSS", "MSI-High"))
circ_data$BRAF.V600E <- factor(circ_data$BRAF.V600E, levels = c("WT", "MUT"), labels = c("Wild-Type", "V600E"))
circ_data$RAS <- factor(circ_data$RAS, levels = c("WT", "MUT"), labels = c("Wild-Type", "Mutant"))
surv_object <- Surv(time = circ_data$DFS.months, event = circ_data$DFS.Event) 
cox_fit <- coxph(surv_object ~ ctDNA.Surveillance + Gender + Age.Group + PrimSite + ECOG + pT + pN + MSI + BRAF.V600E + RAS, data=circ_data) 
ggforest(cox_fit, data = circ_data, main = "Multivariate Regression Model for DFS - All Stages", refLabel = "Reference Group")

test.ph <- cox.zph(cox_fit)

#OS by ctDNA at the MRD Window - pts with Radiological Recurrence

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$RFS.Event=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   1 observation deleted due to missingness 
                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 219     22     NA      NA      NA
ctDNA.MRD=POSITIVE 263     52   43.4    36.8      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$OS.MRD.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="OS - Radiological Recurrence | ctDNA MRD window", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24, 36))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

1 observation deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24    110      12    0.926  0.0209        0.873        0.958
   36     21       9    0.830  0.0364        0.744        0.889

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24     84      37    0.783  0.0334        0.708        0.840
   36     13      13    0.626  0.0490        0.522        0.714

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 482, number of events= 74 
   (1 observation deleted due to missingness)

                    coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE 0.9954    2.7059   0.2557 3.893 9.89e-05 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     2.706     0.3696     1.639     4.466

Concordance= 0.631  (se = 0.027 )
Likelihood ratio test= 16.67  on 1 df,   p=4e-05
Wald test            = 15.16  on 1 df,   p=1e-04
Score (logrank) test = 16.43  on 1 df,   p=5e-05

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 2.71 (1.64-4.47); p = 0"

#OS by ctDNA at the MRD Window - pts with Radiological Recurrence Sites

# Define the function to analyze each recurrence site and extract HR values
analyze_site <- function(site) {
  circ_data_site <- circ_data %>% filter(grepl(site, RelSite, ignore.case = TRUE))
  circ_data_site <- circ_data_site[circ_data_site$ctDNA.MRD != "",]
circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
  
  surv_object <- Surv(time = circ_data_site$OS.MRD.months, event = circ_data_site$OS.Event)
  cox_fit <- coxph(surv_object ~ ctDNA.MRD, data = circ_data_site) 
  cox_fit_summary <- summary(cox_fit)
  
  HR <- cox_fit_summary$coefficients[2]
  lower_CI <- cox_fit_summary$conf.int[3]
  upper_CI <- cox_fit_summary$conf.int[4]
  p_value <- cox_fit_summary$coefficients[5]
  
  label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", format.pval(p_value, digits = 3))
  return(list(HR = HR, lower_CI = lower_CI, upper_CI = upper_CI, p_value = p_value, site = site, label_text = label_text))
}

setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible == "TRUE",]
circ_data <- circ_data[circ_data$RFS.Event == "TRUE",]
recurrence_sites <- c("liver", "lung", "peritoneum", "lymph node")
results <- lapply(recurrence_sites, analyze_site)
forest_data <- do.call(rbind, lapply(results, function(res) {
  data.frame(
    site = res$site,
    HR = res$HR,
    lower_CI = res$lower_CI,
    upper_CI = res$upper_CI,
    label_text = res$label_text
  )
}))

forest_data$site <- factor(forest_data$site, levels = c("liver", "lung", "peritoneum", "lymph node"))
forest_plot <- ggplot(forest_data, aes(x = site, y = HR, ymin = lower_CI, ymax = upper_CI)) +
  geom_pointrange() +
  geom_text(aes(label = label_text), hjust = -0.1, vjust = -0.5) +
  geom_hline(yintercept = 1, linetype = "dashed") +
  coord_flip() +
  scale_y_continuous(breaks = seq(1, max(forest_data$upper_CI) + 1, by = 2), expand = c(0, 0), limits = c(0, max(forest_data$upper_CI) + 1)) +
  labs(x = "Recurrence Site", y = "HR for OS between ctDNA MRD positive vs negative") +
  theme_minimal()
# Define the function to analyze each recurrence site and extract HR values
analyze_site <- function(site) {
  circ_data_site <- circ_data %>% filter(grepl(site, RelSite, ignore.case = TRUE))
  circ_data_site <- circ_data_site[circ_data_site$ctDNA.MRD != "",]
circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
  
  surv_object <- Surv(time = circ_data_site$OS.months, event = circ_data_site$OS.Event)
  cox_fit <- coxph(surv_object ~ ctDNA.MRD, data = circ_data_site) 
  cox_fit_summary <- summary(cox_fit)
  
  HR <- cox_fit_summary$coefficients[2]
  lower_CI <- cox_fit_summary$conf.int[3]
  upper_CI <- cox_fit_summary$conf.int[4]
  p_value <- cox_fit_summary$coefficients[5]
  
  label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", format.pval(p_value, digits = 3))
  return(list(HR = HR, lower_CI = lower_CI, upper_CI = upper_CI, p_value = p_value, site = site, label_text = label_text))
}

# Set working directory and load data
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible == "TRUE",]
circ_data <- circ_data[circ_data$RFS.Event == "TRUE",]
circ_data <- circ_data[circ_data$OS.MRD.months>=0,]

# Recurrence sites to analyze
recurrence_sites <- c("liver", "lung", "peritoneum", "lymph node")

# Perform analysis for each site
results <- lapply(recurrence_sites, analyze_site)

# Create data frame for forest plot
forest_data <- do.call(rbind, lapply(results, function(res) {
  data.frame(
    site = res$site,
    HR = res$HR,
    lower_CI = res$lower_CI,
    upper_CI = res$upper_CI,
    label_text = res$label_text
  )
}))

# Set the order of the levels for the 'site' factor
forest_data$site <- factor(forest_data$site, levels = c("liver", "lung", "peritoneum", "lymph node"))

# Create forest plot
forest_plot <- ggplot(forest_data, aes(x = site, y = HR, ymin = lower_CI, ymax = upper_CI)) +
  geom_pointrange() +
  geom_text(aes(label = label_text), hjust = -0.1, vjust = -0.5) +
  geom_hline(yintercept = 1, linetype = "dashed") +
  coord_flip() +
  scale_y_continuous(breaks = seq(1, max(forest_data$upper_CI) + 1, by = 2), expand = c(0, 0), limits = c(0, max(forest_data$upper_CI) + 1)) +
  labs(x = "Recurrence Site", y = "HR for OS between ctDNA MRD positive vs negative") +
  theme_minimal()

print(forest_plot)

for (res in results) {
  print(res$label_text)
}

[1] "HR = 2.43 (1.01-5.86); p = 0.048"
[1] "HR = 2.64 (1.2-5.83); p = 0.016"
[1] "HR = 2.73 (1.31-5.7); p = 0.007"
[1] "HR = 2.67 (0.83-8.55); p = 0.098"

#OS by ctDNA at the Surveillance Window - pts with Radiological Recurrence

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$RFS.Event=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.Surveillance!="",]
circ_data$OS.months=circ_data$OS.months-2.5
circ_data <- circ_data[circ_data$OS.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$OS.months, event = circ_data$OS.Event)~ctDNA.Surveillance, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$OS.months, event = circ_data$OS.Event) ~ 
    ctDNA.Surveillance, data = circ_data)

                              n events median 0.95LCL 0.95UCL
ctDNA.Surveillance=NEGATIVE  78      2     NA      NA      NA
ctDNA.Surveillance=POSITIVE 264     41   41.8    37.3      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Surveillance) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$OS.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Surveillance, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="OS - Radiological Recurrence | ctDNA Surveillance window", ylab= "Overall Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24, 36))

Call: survfit(formula = surv_object ~ ctDNA.Surveillance, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.Surveillance=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24     48       0    1.000  0.0000           NA           NA
   36      3       2    0.931  0.0471        0.751        0.982

                ctDNA.Surveillance=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
   24     90      31    0.809  0.0325        0.736        0.864
   36     14       8    0.680  0.0592        0.548        0.780

circ_data$ctDNA.Surveillance <- factor(circ_data$ctDNA.Surveillance, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.Surveillance, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Surveillance, data = circ_data)

  n= 342, number of events= 43 

                             coef exp(coef) se(coef)     z Pr(>|z|)   
ctDNA.SurveillancePOSITIVE 2.1278    8.3962   0.7252 2.934  0.00334 **
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.SurveillancePOSITIVE     8.396     0.1191     2.027     34.78

Concordance= 0.631  (se = 0.015 )
Likelihood ratio test= 16.74  on 1 df,   p=4e-05
Wald test            = 8.61  on 1 df,   p=0.003
Score (logrank) test = 12.36  on 1 df,   p=4e-04

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 8.4 (2.03-34.78); p = 0.003"

#Percentage of ctDNA MRD Window positivity in pts undergoing post-recurrence curative surgery

rm(list = ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data %>%
  filter(Eligible == "TRUE" & RFS.Event == "TRUE" & ctDNA.MRD != "")
circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels = c("NEGATIVE", "POSITIVE"), labels = c("Negative", "Positive"))
positive_rate <- sum(circ_data$ctDNA.MRD == "Positive" & circ_data$PostRecurrenceSurgery == "TRUE") / sum(circ_data$ctDNA.MRD == "Positive")* 100
positive_ci <- binconf(sum(circ_data$ctDNA.MRD == "Positive" & circ_data$PostRecurrenceSurgery == "TRUE"),
                       sum(circ_data$ctDNA.MRD == "Positive"),
                       alpha = 0.05)[c(2, 3)] * 100
negative_rate <- sum(circ_data$ctDNA.MRD == "Negative" & circ_data$PostRecurrenceSurgery == "TRUE") / sum(circ_data$ctDNA.MRD == "Negative")* 100
negative_ci <-  binconf(sum(circ_data$ctDNA.MRD == "Negative" & circ_data$PostRecurrenceSurgery == "TRUE"),
                        sum(circ_data$ctDNA.MRD == "Negative"),
                        alpha = 0.05)[c(2, 3)] * 100
data <- data.frame(
  ctDNA.MRD = c("Positive", "Negative"),
  percentage = c(positive_rate, negative_rate),
  lower_ci = c(positive_ci[1], negative_ci[1]),
  upper_ci = c(positive_ci[2], negative_ci[2])
)
cross_tab <- table(circ_data$ctDNA.MRD, circ_data$PostRecurrenceSurgery)
chi_test <- chisq.test(cross_tab)
p_value <- format.pval(chi_test$p.value, digits = 3)
print(data)

print(cross_tab)

          
           FALSE TRUE
  Negative   129   90
  Positive   185   79

print(chi_test)

    Pearson's Chi-squared test with Yates' continuity correction

data:  cross_tab
X-squared = 6.0858, df = 1, p-value = 0.01363

barplot <- ggplot(data, aes(x = ctDNA.MRD, y = percentage, fill = ctDNA.MRD)) +
  geom_bar(stat = "identity") +
  geom_errorbar(aes(ymin = lower_ci, ymax = upper_ci), width = 0.2) +
  geom_text(aes(label = paste0(round(percentage, 1), "%")), vjust = -0.5) +
  labs(
    x = "ctDNA status at the MRD status",
    y = "Proportion of patients undergoing 
    post-recurrence curative surgery",
    caption = paste("Chi-squared test p-value: ", p_value)
  ) +
  scale_y_continuous(expand = c(0, 0), limits = c(0, 50)) +
  scale_fill_manual(values = c("Negative" = "blue", "Positive" = "red")) +
  theme_minimal()
print(barplot)

#PRS by ctDNA at the MRD Window - pts with Radiological Recurrence

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$RFS.Event=="TRUE",]
circ_data <- circ_data[circ_data$OS.MRD.months>=0,]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]

survfit(Surv(time = circ_data$PRS.months, event = circ_data$OS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$PRS.months, event = circ_data$OS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   18 observations deleted due to missingness 
                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 219     22     NA    36.3      NA
ctDNA.MRD=POSITIVE 263     52   38.2    29.2      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$PRS.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="PRS - Radiological Recurrence | ctDNA MRD window", ylab= "Post-Recurrence Survival", xlab="Time from Radiological Recurrence (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

18 observations deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      38.0000      21.0000       0.8073       0.0412       0.7105       0.8745 

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      49.0000      45.0000       0.6809       0.0435       0.5872       0.7577 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 482, number of events= 74 
   (18 observations deleted due to missingness)

                    coef exp(coef) se(coef)    z Pr(>|z|)   
ctDNA.MRDPOSITIVE 0.6772    1.9683   0.2546 2.66  0.00782 **
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     1.968     0.5081     1.195     3.242

Concordance= 0.579  (se = 0.03 )
Likelihood ratio test= 7.63  on 1 df,   p=0.006
Wald test            = 7.08  on 1 df,   p=0.008
Score (logrank) test = 7.35  on 1 df,   p=0.007

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 1.97 (1.2-3.24); p = 0.008"

#PRS by ctDNA at the Surveillance Window - pts with Radiological Recurrence

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$RFS.Event=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.Surveillance!="",]

survfit(Surv(time = circ_data$PRS.months, event = circ_data$OS.Event)~ctDNA.Surveillance, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$PRS.months, event = circ_data$OS.Event) ~ 
    ctDNA.Surveillance, data = circ_data)

                              n events median 0.95LCL 0.95UCL
ctDNA.Surveillance=NEGATIVE  78      2     NA      NA      NA
ctDNA.Surveillance=POSITIVE 264     41   38.2    36.3      NA

event_summary <- circ_data %>%
  group_by(ctDNA.Surveillance) %>%
  summarise(
    Total = n(),
    Events = sum(OS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$PRS.months, event = circ_data$OS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.Surveillance, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="PRS - Radiological Recurrence | ctDNA Surveillance window", ylab= "Post-Recurrence Survival", xlab="Time from Radiological Recurrence (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(24))

Call: survfit(formula = surv_object ~ ctDNA.Surveillance, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.Surveillance=NEGATIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      11.0000       2.0000       0.9317       0.0511       0.7237       0.9847 

                ctDNA.Surveillance=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
      24.000       41.000       38.000        0.700        0.045        0.602        0.778 

circ_data$ctDNA.Surveillance <- factor(circ_data$ctDNA.Surveillance, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.Surveillance, data=circ_data) 
summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.Surveillance, data = circ_data)

  n= 342, number of events= 43 

                             coef exp(coef) se(coef)     z Pr(>|z|)   
ctDNA.SurveillancePOSITIVE 1.8831    6.5739   0.7248 2.598  0.00938 **
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                           exp(coef) exp(-coef) lower .95 upper .95
ctDNA.SurveillancePOSITIVE     6.574     0.1521     1.588     27.21

Concordance= 0.606  (se = 0.02 )
Likelihood ratio test= 12.21  on 1 df,   p=5e-04
Wald test            = 6.75  on 1 df,   p=0.009
Score (logrank) test = 8.99  on 1 df,   p=0.003

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 6.57 (1.59-27.21); p = 0.009"

#Detection ctDNA rates based on sites of relapse

# Remove existing objects and set the working directory
rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$RFS.Event=="TRUE",]

# Create a table of counts for the "Rec.Site" variable
relsite_counts <- table(circ_data$Rec.Site)
relsite_df <- as.data.frame(relsite_counts)
names(relsite_df) <- c("RelSite", "Count")
circ_data_pos_mrd <- circ_data[circ_data$ctDNA.MRD=="POSITIVE",]
circ_data_pos_anytime <- circ_data[circ_data$ctDNA.anytime=="POSITIVE",]
pos_counts_mrd <- table(circ_data_pos_mrd$Rec.Site)
pos_counts_anytime <- table(circ_data_pos_anytime$Rec.Site)
relsite_df$MRDPos_Count <- ifelse(is.na(match(relsite_df$RelSite, names(pos_counts_mrd))), 0, pos_counts_mrd[match(relsite_df$RelSite, names(pos_counts_mrd))])
relsite_df$MRDPos_Count[is.na(relsite_df$MRDPos_Count)] <- 0
relsite_df$AnytimePos_Count <- ifelse(is.na(match(relsite_df$RelSite, names(pos_counts_anytime))), 0, pos_counts_anytime[match(relsite_df$RelSite, names(pos_counts_anytime))])
relsite_df$AnytimePos_Count[is.na(relsite_df$AnytimePos_Count)] <- 0
relsite_df$Percent <- (relsite_df$Count / sum(relsite_df$Count)) * 100
relsite_df$MRDPos_Percent <- (relsite_df$MRDPos_Count / relsite_df$Count) * 100
relsite_df$AnytimePos_Percent <- (relsite_df$AnytimePos_Count / relsite_df$Count) * 100
total_observations <- sum(relsite_df$Count)
total_pos_mrd <- sum(relsite_df$MRDPos_Count)
total_pos_anytime <- sum(relsite_df$AnytimePos_Count)
total_row <- data.frame(RelSite = "Total", Count = total_observations, MRDPos_Count = total_pos_mrd, AnytimePos_Count = total_pos_anytime, Percent = 100, MRDPos_Percent = (total_pos_mrd / total_observations) * 100, AnytimePos_Percent = (total_pos_anytime / total_observations) * 100)
relsite_df <- rbind(relsite_df, total_row)
print(relsite_df)

#Heatmap for Biomarkers factors

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data %>% arrange(RAS.BRAF)
circ_data$RAS <- factor(circ_data$RAS.BRAF, levels = c("TRUE", "FALSE"))
circ_datadf <- as.data.frame(circ_data)

ha <- HeatmapAnnotation(
  RAS.BRAF = circ_data$RAS.BRAF,
  TMB = circ_data$TMB,
  MSI = circ_data$MSI,
  BRAF.V600E = circ_data$BRAF.V600E,
  KRAS.G12C = circ_data$KRAS.G12C,
  ERBB2 = circ_data$ERBB2,
  TP53.Y220C = circ_data$TP53.Y220C,
  NTRK = circ_data$NTRK,
  RET = circ_data$RET,
  
    col = list(RAS.BRAF = c("TRUE" = "blue","FALSE" = "grey"),
    TMB = c("TMB-High" = "blue" , "TMB-Low" = "grey"),
    MSI = c("MSI-High" = "blue" , "MSS" = "grey"),
    BRAF.V600E = c("MUT" = "blue", "WT" = "grey"),
    KRAS.G12C = c("MUT" = "blue", "WT" = "grey"),
    ERBB2 = c("MUT" = "blue", "WT" = "grey"),
    TP53.Y220C = c("MUT" = "blue", "WT" = "grey"),
    NTRK = c("MUT" = "blue", "WT" = "grey"),
    RET = c("MUT" = "blue", "WT" = "grey")))
ht <- Heatmap(matrix(nrow = 0, ncol = length(circ_data$RAS.BRAF)),show_row_names = FALSE,cluster_rows = F,cluster_columns = FALSE, top_annotation = ha)
pdf("heatmap.pdf",width = 7, height = 7)
draw(ht, annotation_legend_side = "bottom")
dev.off()

null device 
          1 

#Calculate the % altered variables

setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible == "TRUE",]
conditions <- list(
  RAS.BRAF = "TRUE",
  TMB = "TMB-High",
  MSI = "MSI-High",
  BRAF.V600E = "MUT",
  KRAS.G12C = "MUT",
  ERBB2 = "MUT",
  TP53.Y220C = "MUT",
  NTRK = "MUT",
  RET = "MUT"
)
total_observations <- nrow(circ_data)
condition_counts <- list()
for (var in names(conditions)) {
  condition_value <- conditions[[var]]
  condition_count <- sum(circ_data[[var]] == condition_value, na.rm = TRUE)
  condition_percentage <- (condition_count / total_observations) * 100
  condition_counts[[var]] <- list('Count' = condition_count, 'Percentage' = condition_percentage)
}
condition_counts_df <- do.call(rbind, lapply(names(condition_counts), function(x) {
  data.frame(Variable = x, 
             Count = condition_counts[[x]]$Count, 
             Percentage = condition_counts[[x]]$Percentage)
}))
print(condition_counts_df)

#DFS by Biomarkers

setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data %>% filter(Eligible == "TRUE")
circ_data <- circ_data %>% 
  mutate(
    RAS.BRAF = ifelse(RAS.BRAF == "TRUE", "RAS/BRAF WT", NA),
    TMB = ifelse(TMB == "TMB-High", "TMB High", NA),
    MSI = ifelse(MSI == "MSI-High", "MSI High", NA),
    BRAF.V600E = ifelse(BRAF.V600E == "MUT", "BRAF V600E", NA),
    KRAS.G12C = ifelse(KRAS.G12C == "MUT", "KRAS G12C", NA),
    ERBB2 = ifelse(ERBB2 == "MUT", "ERBB2", NA),
    TP53.Y220C = ifelse(TP53.Y220C == "MUT", "TP53 Y220C", NA)
  )
circ_data_long <- circ_data %>%
  gather(key = "group", value = "value", RAS.BRAF, TMB, MSI, BRAF.V600E, KRAS.G12C, ERBB2, TP53.Y220C) %>%
  filter(!is.na(value))
circ_data_long$value <- factor(circ_data_long$value, levels = c("RAS/BRAF WT", "TMB High", "MSI High", "BRAF V600E", "KRAS G12C", "ERBB2", "TP53 Y220C"))

survfit(Surv(time = circ_data_long$DFS.months, event = circ_data_long$DFS.Event)~value, data = circ_data_long)

Call: survfit(formula = Surv(time = circ_data_long$DFS.months, event = circ_data_long$DFS.Event) ~ 
    value, data = circ_data_long)

                     n events median 0.95LCL 0.95UCL
value=RAS/BRAF WT 1125    233     NA      NA      NA
value=TMB High     230     10     NA      NA      NA
value=MSI High     215      8     NA      NA      NA
value=BRAF V600E   178     25     NA      NA      NA
value=KRAS G12C     49     19   33.7    22.1      NA
value=ERBB2         36     12     NA    23.2      NA
value=TP53 Y220C    24      6     NA      NA      NA

event_summary <- circ_data_long %>%
  group_by(value) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_obj <- Surv(time = circ_data_long$DFS.months, event = circ_data_long$DFS.Event)
cox_model <- coxph(surv_obj ~ value, data = circ_data_long)
summary(cox_model)

Call:
coxph(formula = surv_obj ~ value, data = circ_data_long)

  n= 1857, number of events= 313 

                   coef exp(coef) se(coef)      z Pr(>|z|)    
valueTMB High   -1.6745    0.1874   0.3230 -5.184 2.17e-07 ***
valueMSI High   -1.8298    0.1605   0.3596 -5.088 3.62e-07 ***
valueBRAF V600E -0.4366    0.6462   0.2105 -2.074  0.03806 *  
valueKRAS G12C   0.7798    2.1810   0.2387  3.267  0.00109 ** 
valueERBB2       0.5571    1.7456   0.2961  1.882  0.05987 .  
valueTP53 Y220C  0.2368    1.2671   0.4135  0.573  0.56693    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                exp(coef) exp(-coef) lower .95 upper .95
valueTMB High      0.1874     5.3362   0.09950    0.3529
valueMSI High      0.1605     6.2324   0.07929    0.3247
valueBRAF V600E    0.6462     1.5474   0.42779    0.9762
valueKRAS G12C     2.1810     0.4585   1.36608    3.4821
valueERBB2         1.7456     0.5729   0.97711    3.1185
valueTP53 Y220C    1.2671     0.7892   0.56344    2.8497

Concordance= 0.635  (se = 0.012 )
Likelihood ratio test= 107  on 6 df,   p=<2e-16
Wald test            = 73.9  on 6 df,   p=6e-14
Score (logrank) test = 93.74  on 6 df,   p=<2e-16

KM_curve <- survfit(surv_obj ~ value, data = circ_data_long)
ggsurvplot(
  KM_curve, 
  data = circ_data_long,
  risk.table = TRUE,
  pval = FALSE,
  conf.int = FALSE,
  break.time.by = 6,
  xlab = "Time from surgery (months)",
  ylab = "Disease-free Survival",
  legend.labs = c("RAS/BRAF WT", "TMB High", "MSI High", "BRAF V600E", "KRAS G12C", "ERBB2", "TP53 Y220C"),
  palette = c("red", "purple", "green", "blue", "orange", "skyblue", "cyan")
)

summary(KM_curve, times = c(24))

Call: survfit(formula = surv_obj ~ value, data = circ_data_long)

                value=RAS/BRAF WT 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     366.0000     224.0000       0.7755       0.0137       0.7491       0.8028 

                value=TMB High 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000     101.0000      10.0000       0.9471       0.0169       0.9146       0.9807 

                value=MSI High 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      96.0000       8.0000       0.9558       0.0159       0.9252       0.9874 

                value=BRAF V600E 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      69.0000      25.0000       0.8382       0.0311       0.7793       0.9015 

                value=KRAS G12C 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       9.0000      18.0000       0.6023       0.0765       0.4696       0.7726 

                value=ERBB2 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000      13.0000      12.0000       0.6287       0.0887       0.4769       0.8289 

                value=TP53 Y220C 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
     24.0000       6.0000       6.0000       0.7237       0.0993       0.5530       0.9470 

#Percentage of ctDNA MRD Window positivity in biomarker groups

setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data %>% filter(Eligible == "TRUE")
circ_data <- circ_data %>% 
  mutate(
    RAS.BRAF = ifelse(RAS.BRAF == "TRUE", "RAS/BRAF WT", NA),
    TMB = ifelse(TMB == "TMB-High", "TMB High", NA),
    MSI = ifelse(MSI == "MSI-High", "MSI High", NA),
    BRAF.V600E = ifelse(BRAF.V600E == "MUT", "BRAF V600E", NA),
    KRAS.G12C = ifelse(KRAS.G12C == "MUT", "KRAS G12C", NA),
    ERBB2 = ifelse(ERBB2 == "MUT", "ERBB2", NA),
    TP53.Y220C = ifelse(TP53.Y220C == "MUT", "TP53 Y220C", NA)
  )
circ_data_long <- circ_data %>%
  gather(key = "group", value = "value", RAS.BRAF, TMB, MSI, BRAF.V600E, KRAS.G12C, ERBB2, TP53.Y220C) %>%
  filter(!is.na(value))

summary_data <- circ_data_long %>%
  group_by(value) %>%
  summarise(
    n = n(),
    positive = sum(ctDNA.MRD == "POSITIVE"),
    pct_positive = (positive / n) * 100,
    se = sqrt((pct_positive / 100) * (1 - pct_positive / 100) / n),
    ci_low = pct_positive - 1.96 * se * 100,
    ci_high = pct_positive + 1.96 * se * 100
  )

overall_summary <- circ_data_long %>%
  summarise(
    value = "Overall",
    n = n(),
    positive = sum(ctDNA.MRD == "POSITIVE"),
    pct_positive = (positive / n) * 100,
    se = sqrt((pct_positive / 100) * (1 - pct_positive / 100) / n),
    ci_low = pct_positive - 1.96 * se * 100,
    ci_high = pct_positive + 1.96 * se * 100
  )

summary_data <- bind_rows(overall_summary, summary_data)

summary_data$value <- factor(summary_data$value, levels = c("Overall", "RAS/BRAF WT", "TMB High", "MSI High", "BRAF V600E", "KRAS G12C", "ERBB2", "TP53 Y220C"))
ggplot(summary_data, aes(x = value, y = pct_positive)) +
  geom_bar(stat = "identity", fill = "blue", alpha = 0.7) +
  geom_errorbar(aes(ymin = ci_low, ymax = ci_high), width = 0.2) +
  geom_text(aes(label = sprintf("%.1f%%", pct_positive)), vjust = -0.5, color = "black") +
  labs(
    x = "Genetic Mutation",
    y = "Post-surgical MRD positivity %"
  ) +
  theme(
    panel.background = element_blank(),
    panel.grid.major = element_blank(),
    panel.grid.minor = element_blank(),
    axis.line = element_line(color = "black"),
    axis.ticks = element_line(color = "black"),
    axis.text.x = element_text(angle = 45, hjust = 1),
    plot.background = element_blank())

#DFS by ctDNA at the MRD Window - All pts Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

   1 observation deleted due to missingness 
                      n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 1773    233     NA      NA      NA
ctDNA.MRD=POSITIVE  336    263   5.34    4.83     6.7

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | All Stages", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

1 observation deleted due to missingness 
                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival  std.err lower 95% CI upper 95% CI
    0   1773       2    0.999 0.000797        0.995        1.000
   24    625     222    0.851 0.009494        0.832        0.869

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    336       4    0.988 0.00592        0.969        0.996
   24     36     254    0.206 0.02364        0.161        0.254

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 2109, number of events= 496 
   (1 observation deleted due to missingness)

                      coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  2.48392  11.98819  0.09162 27.11   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     11.99    0.08342     10.02     14.35

Concordance= 0.738  (se = 0.01 )
Likelihood ratio test= 631.6  on 1 df,   p=<2e-16
Wald test            = 734.9  on 1 df,   p=<2e-16
Score (logrank) test = 1164  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 11.99 (10.02-14.35); p = 0"

#DFS by ctDNA at the MRD Window - RAS/BRAF WT Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$RAS.BRAF=="TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 891    101     NA      NA      NA
ctDNA.MRD=POSITIVE 166    123   6.37    5.06    10.3

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | RAS/BRAF WT", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    891       2    0.998 0.00159        0.991        0.999
   24    316      94    0.873 0.01258        0.846        0.895

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    166       2    0.988 0.00847        0.953        0.997
   24     22     118    0.244 0.03605        0.177        0.317

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 1057, number of events= 224 

                     coef exp(coef) se(coef)    z Pr(>|z|)    
ctDNA.MRDPOSITIVE  2.4770   11.9052   0.1361 18.2   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     11.91      0.084     9.118     15.54

Concordance= 0.742  (se = 0.015 )
Likelihood ratio test= 291  on 1 df,   p=<2e-16
Wald test            = 331.4  on 1 df,   p=<2e-16
Score (logrank) test = 527.1  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 11.91 (9.12-15.54); p = 0"

#DFS by ctDNA at the MRD Window - TMB High Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$TMB=="TMB-High",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 210      5     NA      NA      NA
ctDNA.MRD=POSITIVE   7      4   4.73   0.559      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | TMB-High", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    210       0    1.000  0.0000        1.000        1.000
   24     90       5    0.966  0.0155        0.917        0.986

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0      7       0    1.000   0.000       1.0000        1.000
   24      3       4    0.429   0.187       0.0978        0.734

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 217, number of events= 9 

                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  3.5631   35.2728   0.6756 5.274 1.33e-07 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     35.27    0.02835     9.384     132.6

Concordance= 0.755  (se = 0.083 )
Likelihood ratio test= 18.23  on 1 df,   p=2e-05
Wald test            = 27.82  on 1 df,   p=1e-07
Score (logrank) test = 72.18  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 35.27 (9.38-132.58); p = 0"

#DFS by ctDNA at the MRD Window - MSI High Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$MSI=="MSI-High",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 196      3     NA      NA      NA
ctDNA.MRD=POSITIVE   6      4   2.68   0.559      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | MSI-High", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    196       0    1.000  0.0000        1.000        1.000
   24     86       3    0.977  0.0136        0.928        0.993

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0      6       0    1.000   0.000       1.0000        1.000
   24      2       4    0.333   0.192       0.0461        0.676

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 202, number of events= 7 

                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  4.2671   71.3153   0.7729 5.521 3.37e-08 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     71.32    0.01402     15.68     324.4

Concordance= 0.822  (se = 0.086 )
Likelihood ratio test= 22.32  on 1 df,   p=2e-06
Wald test            = 30.48  on 1 df,   p=3e-08
Score (logrank) test = 112.6  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 71.32 (15.68-324.37); p = 0"

#DFS by ctDNA at the MRD Window - BRAF V600E Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$BRAF.V600E=="MUT",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                     n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 152     12     NA      NA      NA
ctDNA.MRD=POSITIVE  11     11   2.89    1.38      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | BRAF V600E", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    152       0    1.000  0.0000        1.000        1.000
   24     65      12    0.897  0.0296        0.821        0.942

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
           0           11            0            1            0            1            1 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 163, number of events= 23 

                      coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE   5.5020  245.1912   0.8061 6.826 8.75e-12 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     245.2   0.004078     50.51      1190

Concordance= 0.764  (se = 0.049 )
Likelihood ratio test= 67.61  on 1 df,   p=<2e-16
Wald test            = 46.59  on 1 df,   p=9e-12
Score (logrank) test = 265.5  on 1 df,   p=<2e-16

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 245.19 (50.51-1190.25); p = 0"

#DFS by ctDNA at the MRD Window - KRAS G12C Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$KRAS.G12C=="MUT",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                    n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 36      8     NA      NA      NA
ctDNA.MRD=POSITIVE 11     10   2.14    1.61      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | KRAS G12C", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0     36       0    1.000  0.0000        1.000        1.000
   24      8       8    0.759  0.0756        0.572        0.873

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
           0           11            0            1            0            1            1 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 47, number of events= 18 

                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  2.3978   10.9994   0.4904 4.889 1.01e-06 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE        11    0.09091     4.206     28.76

Concordance= 0.746  (se = 0.048 )
Likelihood ratio test= 21.56  on 1 df,   p=3e-06
Wald test            = 23.9  on 1 df,   p=1e-06
Score (logrank) test = 35.4  on 1 df,   p=3e-09

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 11 (4.21-28.76); p = 0"

#DFS by ctDNA at the MRD Window - ERBB2 Amplification Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$ERBB2=="MUT",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                    n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 23      3     NA      NA      NA
ctDNA.MRD=POSITIVE 10      9    4.8    1.84      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | ERBB2 Amplification", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0     23       0    1.000  0.0000        1.000        1.000
   24     11       3    0.855  0.0778        0.613        0.951

                ctDNA.MRD=POSITIVE 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
           0           10            0            1            0            1            1 

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 33, number of events= 12 

                     coef exp(coef) se(coef)     z Pr(>|z|)    
ctDNA.MRDPOSITIVE  2.8717   17.6668   0.6922 4.148 3.35e-05 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     17.67     0.0566     4.549     68.61

Concordance= 0.808  (se = 0.046 )
Likelihood ratio test= 20.56  on 1 df,   p=6e-06
Wald test            = 17.21  on 1 df,   p=3e-05
Score (logrank) test = 29.02  on 1 df,   p=7e-08

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 17.67 (4.55-68.61); p = 0"

#DFS by ctDNA at the MRD Window - TP53 Y220C Landmark MRD timepoint

rm(list=ls())
setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible=="TRUE",]
circ_data <- circ_data[circ_data$TP53.Y220C=="MUT",]
circ_data <- circ_data[circ_data$ctDNA.MRD!="",]
circ_data <- circ_data[circ_data$DFS.MRD.months>=0,]
circ_datadf <- as.data.frame(circ_data)

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~ctDNA.MRD, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    ctDNA.MRD, data = circ_data)

                    n events median 0.95LCL 0.95UCL
ctDNA.MRD=NEGATIVE 18      3     NA      NA      NA
ctDNA.MRD=POSITIVE  4      2   5.39    1.77      NA

event_summary <- circ_data %>%
  group_by(ctDNA.MRD) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <-Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ ctDNA.MRD, data = circ_data,conf.int=0.95,conf.type="log-log") 
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, break.time.by=6, palette=c("blue","red"), title="DFS - ctDNA MRD window | TP53 Y220C", ylab= "Disease-Free Survival", xlab="Time from Landmark Time point (Months)", legend.labs=c("ctDNA Negative", "ctDNA Positive"), legend.title="")

summary(KM_curve, times= c(0, 24))

Call: survfit(formula = surv_object ~ ctDNA.MRD, data = circ_data, 
    conf.int = 0.95, conf.type = "log-log")

                ctDNA.MRD=NEGATIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0     18       0    1.000   0.000        1.000         1.00
   24      5       3    0.787   0.115        0.453         0.93

                ctDNA.MRD=POSITIVE 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0      4       0      1.0    0.00       1.0000        1.000
   24      1       2      0.5    0.25       0.0578        0.845

circ_data$ctDNA.MRD <- factor(circ_data$ctDNA.MRD, levels=c("NEGATIVE","POSITIVE"))
cox_fit <- coxph(surv_object ~ ctDNA.MRD, data=circ_data) 
ggforest(cox_fit,data = circ_data)

summary(cox_fit)

Call:
coxph(formula = surv_object ~ ctDNA.MRD, data = circ_data)

  n= 22, number of events= 5 

                    coef exp(coef) se(coef)    z Pr(>|z|)
ctDNA.MRDPOSITIVE 1.3124    3.7152   0.9177 1.43    0.153

                  exp(coef) exp(-coef) lower .95 upper .95
ctDNA.MRDPOSITIVE     3.715     0.2692    0.6149     22.45

Concordance= 0.661  (se = 0.115 )
Likelihood ratio test= 1.78  on 1 df,   p=0.2
Wald test            = 2.05  on 1 df,   p=0.2
Score (logrank) test = 2.35  on 1 df,   p=0.1

cox_fit_summary <- summary(cox_fit)

# Extract values for HR, 95% CI, and p-value
HR <- cox_fit_summary$coefficients[2]
lower_CI <- cox_fit_summary$conf.int[3]
upper_CI <- cox_fit_summary$conf.int[4]
p_value <- cox_fit_summary$coefficients[5]
label_text <- paste0("HR = ", round(HR, 2), " (", round(lower_CI, 2), "-", round(upper_CI, 2), "); p = ", round(p_value, 3))
print(label_text)

[1] "HR = 3.72 (0.61-22.45); p = 0.153"

#DFS by ctDNA at the MRD Window - Forest plot with all subgroups of biomarkers

setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible == "TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD != "",]
circ_data <- circ_data[circ_data$DFS.MRD.months >= 0,]
perform_cox <- function(data, filter_col = NULL, filter_val = NULL) {
  if (!is.null(filter_col) & !is.null(filter_val)) {
    data <- data[data[[filter_col]] == filter_val,]
  }
  surv_object <- Surv(time = data$DFS.MRD.months, event = data$DFS.Event)
  cox_fit <- coxph(surv_object ~ ctDNA.MRD, data = data)
  cox_fit_summary <- summary(cox_fit)
  HR <- cox_fit_summary$coefficients[2]
  lower_CI <- cox_fit_summary$conf.int[3]
  upper_CI <- cox_fit_summary$conf.int[4]
  p_value <- cox_fit_summary$coefficients[5]
  return(c(HR, lower_CI, upper_CI, p_value))
}

results <- data.frame(
  Subgroup = c("All", "RAS/BRAF WT", "TMB-High", "MSI-High", "BRAF V600E", "KRAS G12C", "ERBB2", "TP53 Y220C"),
  HR = rep(NA, 8),
  lower_CI = rep(NA, 8),
  upper_CI = rep(NA, 8),
  p_value = rep(NA, 8)
)

results[1, 2:5] <- perform_cox(circ_data)
results[2, 2:5] <- perform_cox(circ_data, "RAS.BRAF", "TRUE")
results[3, 2:5] <- perform_cox(circ_data, "TMB", "TMB-High")
results[4, 2:5] <- perform_cox(circ_data, "MSI", "MSI-High")
results[5, 2:5] <- perform_cox(circ_data, "BRAF.V600E", "MUT")
results[6, 2:5] <- perform_cox(circ_data, "KRAS.G12C", "MUT")
results[7, 2:5] <- perform_cox(circ_data, "ERBB2", "MUT")
results[8, 2:5] <- perform_cox(circ_data, "TP53.Y220C", "MUT")

results$HR <- as.numeric(results$HR)
results$lower_CI <- as.numeric(results$lower_CI)
results$upper_CI <- as.numeric(results$upper_CI)
results$p_value <- as.numeric(results$p_value)
results$label_text <- paste0(
  "HR = ", round(results$HR, 2), 
  "\n95% CI = ", round(results$lower_CI, 2), "-", round(results$upper_CI, 2),
  "\np = ", round(results$p_value, 3)
)
ggplot(results, aes(x = Subgroup, y = HR)) +
  geom_point(size = 3) +
  geom_errorbar(aes(ymin = lower_CI, ymax = upper_CI), width = 0.2) +
  geom_text(aes(label = label_text), hjust = -0.2, vjust = 0.5, size = 3.5) +
  scale_y_log10() +
  geom_hline(yintercept = 1, linetype = "dashed") +
  labs(title = "Forest Plot of HR for DFS between ctDNA Positive versus Negative",
       x = "Subgroup",
       y = "Hazard Ratio (HR)") +
  coord_flip() +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))

#DFS by BRAF & MSI - ctDNA Positive Landmark MRD timepoint

setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible == "TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD != "",]
circ_data <- circ_data[circ_data$ctDNA.MRD == "POSITIVE",]
circ_data <- circ_data[circ_data$DFS.MRD.months >= 0,]

# Create the BRAF.MSI variable
circ_data$BRAF.MSI <- NA
circ_data <- circ_data %>%
  mutate(BRAF.MSI = case_when(
    BRAF.V600E == "WT" & MSI == "MSS" ~ 1,
    BRAF.V600E == "WT" & MSI == "MSI-High" ~ 2,
    BRAF.V600E == "MUT" & MSI == "MSI-High" ~ 3,
    BRAF.V600E == "MUT" & MSI == "MSS" ~ 4
  ))

circ_data$BRAF.MSI <- factor(circ_data$BRAF.MSI, levels = c(1, 2, 3, 4), 
                             labels = c("BRAF WT & MSS", "BRAF WT & MSI-High", 
                                        "BRAF V600E & MSI-High", "BRAF V600E & MSS"))

print(table(circ_data$BRAF.MSI, useNA = "ifany"))

        BRAF WT & MSS    BRAF WT & MSI-High BRAF V600E & MSI-High      BRAF V600E & MSS                  <NA> 
                  320                     5                     1                    10                     1 

circ_data <- circ_data[!is.na(circ_data$BRAF.MSI),]
if(nrow(circ_data) == 0) {
  stop("No non-missing observations in the dataset after filtering.")
}
survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~BRAF.MSI, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    BRAF.MSI, data = circ_data)

                                 n events median 0.95LCL 0.95UCL
BRAF.MSI=BRAF WT & MSS         320    249  5.520   4.895    7.16
BRAF.MSI=BRAF WT & MSI-High      5      3  4.731   0.559      NA
BRAF.MSI=BRAF V600E & MSI-High   1      1  0.624      NA      NA
BRAF.MSI=BRAF V600E & MSS       10     10  3.285   1.380      NA

event_summary <- circ_data %>%
  group_by(BRAF.MSI) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <- Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ BRAF.MSI, data = circ_data, conf.int = 0.95, conf.type = "log-log")

# Plot the Kaplan-Meier curve
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, 
           break.time.by = 6, palette = c("blue", "green", "purple", "red"), 
           title = "DFS - BRAF & MSI | ctDNA MRD Positive", ylab = "Disease-Free Survival", 
           xlab = "Time from Landmark Time point (Months)", 
           legend.labs = c("BRAF WT & MSS", "BRAF WT & MSI-High", 
                           "BRAF V600E & MSI-High", "BRAF V600E & MSS"), 
           legend.title = "")

summary(KM_curve, times = c(0, 24))

Call: survfit(formula = surv_object ~ BRAF.MSI, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                BRAF.MSI=BRAF WT & MSS 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    320       4    0.988 0.00621        0.967        0.995
   24     34     240    0.209 0.02448        0.163        0.259

                BRAF.MSI=BRAF WT & MSI-High 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0      5       0      1.0   0.000        1.000        1.000
   24      2       3      0.4   0.219        0.052        0.753

                BRAF.MSI=BRAF V600E & MSI-High 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
           0            1            0            1            0            1            1 

                BRAF.MSI=BRAF V600E & MSS 
        time       n.risk      n.event     survival      std.err lower 95% CI upper 95% CI 
           0           10            0            1            0            1            1 

cox_fit <- coxph(surv_object ~ BRAF.MSI, data = circ_data)
summary(cox_fit)

Call:
coxph(formula = surv_object ~ BRAF.MSI, data = circ_data)

  n= 336, number of events= 263 

                                 coef exp(coef) se(coef)      z Pr(>|z|)   
BRAF.MSIBRAF WT & MSI-High    -0.2883    0.7495   0.5818 -0.496  0.62018   
BRAF.MSIBRAF V600E & MSI-High  2.6324   13.9073   1.0209  2.579  0.00992 **
BRAF.MSIBRAF V600E & MSS       0.7860    2.1947   0.3250  2.419  0.01557 * 
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

                              exp(coef) exp(-coef) lower .95 upper .95
BRAF.MSIBRAF WT & MSI-High       0.7495     1.3342    0.2397     2.344
BRAF.MSIBRAF V600E & MSI-High   13.9073     0.0719    1.8805   102.851
BRAF.MSIBRAF V600E & MSS         2.1947     0.4556    1.1608     4.149

Concordance= 0.511  (se = 0.008 )
Likelihood ratio test= 8.29  on 3 df,   p=0.04
Wald test            = 12.54  on 3 df,   p=0.006
Score (logrank) test = 17.48  on 3 df,   p=6e-04

#DFS by BRAF & MSI - ctDNA Negative Landmark MRD timepoint

setwd("~/Downloads")
circ_data <- read.csv("Galaxy Data_20240603 Complete Dataset.csv")
circ_data <- circ_data[circ_data$Eligible == "TRUE",]
circ_data <- circ_data[circ_data$ctDNA.MRD != "",]
circ_data <- circ_data[circ_data$ctDNA.MRD == "NEGATIVE",]
circ_data <- circ_data[circ_data$DFS.MRD.months > 0,]

circ_data$BRAF.MSI <- NA
circ_data <- circ_data %>%
  mutate(BRAF.MSI = case_when(
    BRAF.V600E == "WT" & MSI == "MSS" ~ 1,
    BRAF.V600E == "WT" & MSI == "MSI-High" ~ 2,
    BRAF.V600E == "MUT" & MSI == "MSI-High" ~ 3,
    BRAF.V600E == "MUT" & MSI == "MSS" ~ 4
  ))

circ_data$BRAF.MSI <- factor(circ_data$BRAF.MSI, levels = c(1, 2, 3, 4), 
                             labels = c("BRAF WT & MSS", "BRAF WT & MSI-High", 
                                        "BRAF V600E & MSI-High", "BRAF V600E & MSS"))
print(table(circ_data$BRAF.MSI, useNA = "ifany"))

        BRAF WT & MSS    BRAF WT & MSI-High BRAF V600E & MSI-High      BRAF V600E & MSS 
                 1526                    93                   103                    49 

circ_data <- circ_data[!is.na(circ_data$BRAF.MSI),]
if (any(!is.finite(circ_data$DFS.MRD.months)) || any(!is.finite(circ_data$DFS.Event))) {
  stop("Data contains non-finite values.")
}
if (nrow(circ_data) == 0) {
  stop("No non-missing observations in the dataset after filtering.")
}

survfit(Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)~BRAF.MSI, data = circ_data)

Call: survfit(formula = Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event) ~ 
    BRAF.MSI, data = circ_data)

                                  n events median 0.95LCL 0.95UCL
BRAF.MSI=BRAF WT & MSS         1526    219     NA      NA      NA
BRAF.MSI=BRAF WT & MSI-High      93      0     NA      NA      NA
BRAF.MSI=BRAF V600E & MSI-High  103      3     NA      NA      NA
BRAF.MSI=BRAF V600E & MSS        49      9     NA      NA      NA

event_summary <- circ_data %>%
  group_by(BRAF.MSI) %>%
  summarise(
    Total = n(),
    Events = sum(DFS.Event),
    Fraction = Events / n(),
    Percentage = (Events / n()) * 100
  )
print(event_summary)

surv_object <- Surv(time = circ_data$DFS.MRD.months, event = circ_data$DFS.Event)
KM_curve <- survfit(surv_object ~ BRAF.MSI, data = circ_data, conf.int = 0.95, conf.type = "log-log")

# Plot the Kaplan-Meier curve
ggsurvplot(KM_curve, data = circ_data, pval = FALSE, conf.int = FALSE, risk.table = TRUE, 
           break.time.by = 6, palette = c("blue", "green", "purple", "red"), 
           title = "DFS - BRAF & MSI | ctDNA MRD Negative", ylab = "Disease-Free Survival", 
           xlab = "Time from Landmark Time point (Months)", 
           legend.labs = c("BRAF WT & MSS", "BRAF WT & MSI-High", 
                           "BRAF V600E & MSI-High", "BRAF V600E & MSS"), 
           legend.title = "")

summary(KM_curve, times = c(0, 24))

Call: survfit(formula = surv_object ~ BRAF.MSI, data = circ_data, conf.int = 0.95, 
    conf.type = "log-log")

                BRAF.MSI=BRAF WT & MSS 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0   1526       0    1.000  0.0000        1.000        1.000
   24    519     210    0.838  0.0106        0.816        0.858

                BRAF.MSI=BRAF WT & MSI-High 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0     93       0        1       0            1            1
   24     41       0        1       0           NA           NA

                BRAF.MSI=BRAF V600E & MSI-High 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0    103       0    1.000  0.0000        1.000        1.000
   24     45       3    0.954  0.0269        0.859        0.985

                BRAF.MSI=BRAF V600E & MSS 
 time n.risk n.event survival std.err lower 95% CI upper 95% CI
    0     49       0    1.000  0.0000        1.000        1.000
   24     20       9    0.788  0.0658        0.622        0.887

cox_fit <- coxphf(surv_object ~ BRAF.MSI, data = circ_data)
summary(cox_fit)

coxphf(formula = surv_object ~ BRAF.MSI, data = circ_data)

Model fitted by Penalized ML
Confidence intervals and p-values by Profile Likelihood 

                                    coef  se(coef)  exp(coef)   lower 0.95 upper 0.95      Chisq            p
BRAF.MSIBRAF WT & MSI-High    -3.4234200 1.4219914 0.03260075 0.0002591613  0.2215540 25.2847247 4.946103e-07
BRAF.MSIBRAF V600E & MSI-High -1.5067027 0.5411239 0.22163959 0.0620647658  0.5473653 13.3792192 2.544276e-04
BRAF.MSIBRAF V600E & MSS       0.2475077 0.3328541 1.28082917 0.6222123751  2.3148209  0.5176429 4.718489e-01

Likelihood ratio test=38.29511 on 3 df, p=2.44771e-08, n=1771
Wald test = 14.17091 on 3 df, p = 0.002681504

Covariance-Matrix:
                              BRAF.MSIBRAF WT & MSI-High BRAF.MSIBRAF V600E & MSI-High BRAF.MSIBRAF V600E & MSS
BRAF.MSIBRAF WT & MSI-High                   2.022059448                   0.004612586              0.004606403
BRAF.MSIBRAF V600E & MSI-High                0.004612586                   0.292815067              0.004594608
BRAF.MSIBRAF V600E & MSS                     0.004606403                   0.004594608              0.110791854